This page intentionally left blank
It is quite commonplace for bilingual speakers to use two or more languages,
dialects or varieties in the same conversation, without any apparent effort. This
phenomenon, known as code-switching, has become a major focus of attention
in linguistics. This concise and original study explores how, when and where
code-switching occurs. Drawing on a diverse range of examples from medieval manuscripts to rap music, novels to advertisements, emails to political
speeches, and above all everyday conversation, it argues that code-switching
can only be properly understood if we study it from a variety of perspectives. It
shows how sociolinguistic, psycholinguistic, grammatical and developmental
aspects of code-switching are all interdependent, and findings in each area are
crucial to others. Breaking down barriers across the discipline of linguistics,
this pioneering book confronts fundamental questions about what a “native
language” is, and whether languages can be meaningfully studied independently from individuals who use them.
p e n e l o p e g a r d n e r - c h l o r o s is Senior Lecturer in the Department of
Applied Linguistics and Communication, Birkbeck, University of London. Her
publications include Language Selection and Switching in Strasbourg (1991) and
Vernacular Literacy: a Re-evaluation (with Tabouret-Keller, Le Page and Varro,
1997). She is a founding member of the LIDES group, which has set up the first
database of bilingual texts. She has published widely on code-switching and on
her other field of interest, Terms of Address, in sociolinguistic journals and
Penelope Gardner-Chloros
Birkbeck College, University of London
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore,
São Paulo, Delhi, Dubai, Tokyo
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
Information on this title: www.cambridge.org/9780521862646
© Penelope Gardner-Chloros 2009
This publication is in copyright. Subject to statutory exception and to the
provision of relevant collective licensing agreements, no reproduction of any part
may take place without the written permission of Cambridge University Press.
First published in print format 2009
eBook (EBL)
Cambridge University Press has no responsibility for the persistence or accuracy
of urls for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
To Alexander, Nicholas, Zoe and Philip
John Godfrey Saxe’s (1816–1887) version of the
famous Indian legend
It was six men of Indostan
To learning much inclined
Who went to see the elephant
(Though all of them were blind),
That each by observation
Might satisfy his mind
The First approached the Elephant
And happening to fall
Against his broad and sturdy side
At once began to bawl:
“God bless me! but the Elephant
Is very like a wall!”
The Second, feeling of the tusk
Cried “Ho! What have we here,
So very round and smooth and sharp
To me tis mighty clear
This wonder of an Elephant
Is very like a spear!”
The Third approached the animal
And happening to take
The squirming trunk within his hands
Thus boldly up he spake:
“I see” quoth he, “the Elephant
Is very like a snake!”
The Fourth reached out an eager hand
And felt about the knee
“What most this wondrous beast is like
Is mighty plain,” quoth he;
“Tis clear enough the Elephant
Is very like a tree!”
The Fifth, who chanced to touch the ear,
Said: “E’en the blindest man
Can tell what this resembles most;
Deny the fact who can,
This marvel of an Elephant
Is very like a fan!”
The Sixth no sooner had begun
About the beast to grope
Than, seizing on the swinging tail
That fell within his scope
“I see,” quoth he, “the Elephant
Is very like a rope!”
And so these men of Indostan
Disputed loud and long
Each in his opinion
Exceeding stiff and strong,
Though each was partly in the right
And all were in the wrong!
This rendition of Saxe’s poem is compiled from two sources: Don Fabun (1968),
Communications, the Transfer of Meaning, New York: Macmillan, p. 13 and John
Godfrey Saxe (1963), The Blind Men and the Elephant; John Godfrey Saxe’s version
of the famous Indian legend. Pictures by Paul Galdone, New York: Whittlesey House;
a letter to me from McGraw-Hill (dated 15 August 1998) states that the text for this
edition appears to be in the public domain, but the illustrations are not. Note that the text
from each of these two sources differs from my version with respect to one line, and they
are different lines. I have not seen the original first edition, but this is my best guess of it.
Incidentally, the original parable originated in China sometime during the Han dynasty
(202 bc–220 ad) as:“Three Blind Men and an Elephant”.
Transcription conventions
Code-switching and language contact
Social factors in code-switching
Code-switching in conversation
Grammatical aspects of code-switching
Psycholinguistic approaches
Acquiring code-switching: code-switching in children
(and L2 learners)
page x
I would like to express my warmest thanks to those – whether named here or not –
who have either helped me, or provided general support and inspiration over the
period of gestation – more elephantine than human – of this book.
Two distinguished scholars of Bilingualism, Andrée Tabouret-Keller and
Michel Blanc, have been a source of ideas, advice, encouragement and friendship throughout the process. This book would not have seen the light without
them. My debt to Bob Le Page, now sadly missed, is also considerable. Together
with Andrée, and inspired by the unfocused linguistic landscapes of Jamaica
and Belize, he is responsible for the fundamental question which underlies
many of the puzzles in this book: “What is a language?” His legacy lives on
in many of today’s most significant linguists. I also owe the greatest thanks to Li
Wei, one of the foremost experts on Bilingualism, who has provided support of
many kinds and commented repeatedly on the manuscript as it developed.
My colleagues at Birkbeck have allowed various periods of leave devoted
to writing this book and provided much fruitful discussion and advice.
Malcolm Edwards gave concrete help beyond the call of duty and deserves
a large part of the credit for Chapter 5, which is partly based on work we did
together; his wicked sense of humour and wit have brightened up the daily
grind for a number of years. Itesh Sachdev, now at SOAS, and Jean-Marc
Dewaele also provided valuable advice and support. Ken Mackley in Birkbeck
Library was a bibliographical hero and showed that humans still have the edge
on Google.
Other colleagues have helped in numerous ways. Apart from working with
me on various aspects of code-switching, Jenny Cheshire has been a first-class
colleague and friend throughout. My LIPPS colleagues have played a major
role, as should be clear from the book: Mark Sebba and Melissa Moyer first and
foremost, but Roeland van Hout, Eva Eppler, Jeanine Treffers-Daller, Pieter
Muysken, Ad Backus, Jacomine Nortier and others involved in the LIPPS
enterprise are all deserving of thanks. Mark Sebba also read and commented
usefully on the manuscript. Lisa McEntee-Atalianis and Katerina Finnis provided active collaboration and assistance but also companionship, without
which doing research would be lonely indeed. For help in sourcing, translating
and understanding the significance of the illustrations, I am most grateful to
Darren Aoki, Tamar Drukker, Tony Hunt, Shahrzad Mahootian, Andrew Mason
and Ian Short.
No book on code-switching can fail to acknowledge a debt to Carol MyersScotton, who has done so much to put code-switching on the map, and who has
always been a generous correspondent. I have also benefited from advice from
Jeff MacSwan. Others who deserve thanks include Peter Auer, Michael Clyne,
Adam Jaworski and François Grosjean. My students at Birkbeck, particularly
on the MA Applied Linguistics and MA Bilingualism, have provided many
insights on, and examples of, code-switching. Needless to say, the responsibility
for all errors, misinterpretations and omissions is mine alone.
My thanks also to Helen Barton and Jodie Barnes at Cambridge University
Press, Adrian Stenton for his editorial work, and the Guggenheim Foundation,
Venice for the right to reproduce the cover picture, especially Mr. Silvio Veronese.
Finally, this book was produced with inspiration – and no small measure of
distraction – from its (plurilingual) dedicatees . My heartfelt thanks go to Piers,
for defending my fundamental human right to be excused, many times, from
domestic duties and to be provided with tea and sympathy as required.
Transcription conventions
(1) Making the examples easy for English-speaking readers to follow has been
given priority over consistency of presentation in different instances, and
the use of non-Latin alphabets has been avoided. Extracts from data
discussed by others have been presented as they originally presented
them, e.g. examples from Auer, Li Wei and Sebba follow the Atkinson
and Heritage (1984) conventions.
(2) A word-for-word (or morpheme-for-morpheme) gloss is sometimes given
below the examples, as well as a free translation below that; at other times
there is a free translation only. This depends partly on what type of point is
being made about the switch, and partly, where the example is taken from
someone else’s work, on whether a gloss was provided in the original
(3) CS has been picked out by the use of bold script for one of the languages
involved (and corresponding use of fonts in the translation below). Where
there is a third language involved this appears in bold italics. The implied
decisions about which words belong to which language are often somewhat
arbitrary and should be taken only as a general indication.
(4) In the examples from data collected in Strasbourg, the spelling of Alsatian is
based on a system derived from German spelling (Matzen, 1980), the
purpose of which is to provide a standardized spelling system for the
Strasbourg dialect.
(5) In Greek/GCD examples, Greek has been transcribed into a semi-phonetic
Roman script, retaining the Greek letters /χ/, /δ/ and /γ/ which sound the
same as those in the IPA (International Phonetic Alphabet). Other sounds
follow English spelling: /θ/ is represented as ‘th’ and /ʃ/ as sh. The phonetic
symbol /j/ is left as an i (e.g. [ja] = ia) so as to avoid confusion with the
English letter j.
What is code-switching?
In Alsace, in eastern France, French is commonly mixed with the local dialect,
Alsatian, which is a variety of German (or more precisely Alemannic). At a family
gathering in Strasbourg on New Year’s Eve, a discussion starts regarding the
poor quality of Alsatian butchers compared to those in the rest of France. One of
the guests, Mr Eder,1 a jovial middle-aged man and a prolific talker, holds forth:
Example 1
1 mr eder: du bekommsch do e fätze … je sais pas dans quelle graisse
you get some sort of scraps … in goodness knows what sort of fat
… avec quoi: avec de de de was weiss denn de teiffel
… with what: with the the the the devil knows what
noh geh i anne un! putz diss ding
then I have to go and clean the thing up
parce que lorsque tu as un morceau de viande im … im teller
because when you have a piece of meat
on … on your plate
un noochher hesch eso gschnuddels un muesch abschniede diss
ganze ding gell.
and then you find you have a sort of mess and you have to cut the
whole thing off you see
oder e so hoch fett uf’m ding … diss haw i halt schliesslich a nitt
gere gell?
or fat this high on top of it … I really don’t like that at all you see
(Gardner-Chloros, 1991:124)
Mr Eder is a fluent speaker of both French and the Alsatian dialect. His apparent
hesitations, represented by dots or repetitions, are found in stretches within the
same language (L2, L4, L6) just as often as between stretches in different
languages – their purpose is dramatic effect. No rhyme or reason appears to
govern the points at which he passes from one language to the other. This form
of expression in bilinguals has been called “mixed discourse” – to say that there
are two separate languages is more or less meaningless from the participants’
point of view. In L2, he also uses a “bridge”, i.e. a word which could come from
All proper names have been changed.
either language, to facilitate the switch: de, which can be a partitive in French or
a definite article in Alsatian.
Example 2
A second generation Greek Cypriot teenager, brought up and living in London,
Olga, told this story about why her father emigrated to England. The interviewer
spoke the Greek Cypriot Dialect so as to encourage Olga to use that variety.
1 interviewer: iati irthan stin Anglia?
why did they come to England?
2 olga:
this is a long story [laughs]. Itun mesa se ena mikro χorio, tin
They were in a little village,
ke o pateras mu – afto mu ipen o pateras mu. δen eksero … an
ine sosto
and my father – this is what my father told me. I don’t know… if it’s
Mu ipe pos mian niχta emethisen ke pien ke epiasen mian…
pos tin lene
he told me.. that one night they got drunk … and they went and
caught a … what’s its name.
5 interviewer: boris
na to pis sta anglika.
You can say it in English.
6 olga:
chicken chicken. Chicken [laughing] pu enan γitonon.…
which (belonged) to a
ke ton ivrasi ke ipen pola. Endrapiken pola.
and they found him and said a lot (had a big discussion). He was
very ashamed.
ke mu lei ia afton ton skopon irten stin Anglia.
and he tells me (that it is) for that purpose [sic] he came to England.
(Gardner-Chloros, unpublished data)
In common with many other second/third generation London Greek Cypriots,
Olga is much more at ease speaking English. She hesitates and searches for her
words when she has to speak Greek. In L2, she plays for time by saying, “This is
a long story” in English. Later, she cannot find the simple everyday word
‘chicken’ in Greek (L4), a fact which is indirectly commented upon by the
interviewer (L5), and there are several other signs of her difficulties with Greek,
such as her omission of the verb ‘belonged’ in L6. Mixing the two languages is
the normal way to talk in her community, but speaking to a purely Greekspeaking interlocutor clearly taxes her competence in Greek.
In Example 3, it may seem to the observer that a single variety is being used,
but those familiar with in-group communication in this community would
recognize that speakers are in fact alternating between different varieties.
This extract is from Sebba’s London Jamaican (1993). Two London teenage
boys of Jamaican parents, Andrew and Barry, are discussing an incident involving another young man, which occurred while Andrew was serving in a shop.
Although Andrew’s entire description may appear to be in a variety of London
English, the passages in bold are in fact in Creole. In some cases, as in the word
Lucozade (L5), it is only the pronunciation which identifies the word as Creole
(and so obviously we are taking the researcher’s word that this is what happens).
In other cases (e.g. the words in bold in L11), other features also tell us that this
is Creole. The passage in L19–20 in capitals is described as being in a near-RP
“posh” voice.2
Example 3
1 andrew: yeah man, I was on the till on Saturday (1.2) and this this black man
come in (1.0) and (0.6) you know our shop, right, (0.6) they u:m (0.2)
give (.) refund on (0.3) Lucozade bottles (0.4)
5 andrew: a black man come in an’ ’im b(h)u::y a bottle (.) of Lucozade while ’e
was in the shop [ an’
[ free p- e’s e got free pee off is it?
andrew: yeah
small ones or big ones?
10 andrew: big ones and ’e drank the bottle in fron% of us an then ask(d) for the
money back (see man) me want me money now
[ heheh
andrew: he goes (pnk) (I’m on) the till guy (.) hhh (I jus) (0.6) I jus’ look round
at ’im (0.6) I said well you can’t ’ave it (1.9)
I said I ’ave to open the till (w) wait till the next customer comes (1.0)
‘now! open it now and give me the money’ (1.0)
I said I can’t (0.8) the man just thump ’is fist down an’ (screw up
dis for me) (.) (s no man) the manager just comes (.)
BEFORE I CALL THE SECURITY: hh the man jus’ take the bottle
an’ fling it at me an’ (I) jus’ catch it at the (ground)
(Sebba, 1993:119–120)
Sebba suggests that code-switching is used here to “animate” the narrative by
providing different “voices” for the participants in the incident which is
described. Although both the customer and the narrator might be expected to
speak the same variety, either London English or Creole, Andrew reserves
Creole mainly to quote the customer and to describe his actions (L17 and 20–21).
These examples show that the behaviour of bilinguals can only be properly
understood with some insider knowledge of the community and the circumstances
RP is “Received Pronunciation”. Pauses are indicated as in the original, as are brackets showing
overlapping speech.
where it is displayed. First, the speakers’ competence in the relevant varieties may
or may not be a determining factor in their choices. Second, an observer may or
may not be able to distinguish which shifts in accent, vocabulary or syntax are in
some way significant for the participants in the conversation.
Why study code-switching?
Such varied combinations of two or more linguistic varieties occur in countless
bilingual societies and communities, and are known as code-switching (CS).3
It refers to the use of several languages or dialects in the same conversation or
sentence by bilingual people. It affects practically everyone who is in contact
with more than one language or dialect, to a greater or lesser extent. Numerous
local names designate such mixed talk: Tex-Mex, Franglais, BBC Grenglish,
Chinglish, Spanglish, Tuti Futi, etc. In some earlier periods of history, CS was
equally common in writing (see the papers in Trotter, 2002). Apart from CS,
there are a number of other possible linguistic outcomes of language contact
including borrowing, convergence, pidginization, language death, etc. CS has
been found to occur alongside most of these, though it does not necessarily do
so. The various manifestations of contact are grouped here under the heading of
language interaction.
Code-switching as a window on speech and language
The study of why and how people code-switch provides insights about many
aspects of language as well as speech. This applies not only to how language
and languages are organized in the brain (the mechanisms of switching as
such are discussed in Chapter 6). At a functional level, bilinguals often switch
varieties in order to communicate something beyond the superficial meaning
of their words. Monolinguals can do this also, by switching between dialects,
registers, levels of formality, intonations etc. (Bell, 1984; Coupland, 1985;
Labov, 1971; Kerswill, 1994).4 “I can do aught when you’re with me, I can
do anything”, said a male speaker in his sixties from Sheffield in a Radio 4
interview, talking to his wife in an aside. Aught is Northern dialect, which
he then repeats in Standard English, anything. Such a switch serves at least
two functions: by using a dialect word, he emphasizes the fact that he is talking
to his wife rather than the interviewer. At the same time, he reinforces his
Code-switching is sometimes found in the literature written as two separate words, sometimes
with a hyphen and sometimes as one word. Diachronially speaking, the move from two words to
hyphenated words to a single word reflects the semantic acceptability and integration of the
concept. I have stuck here with the intermediate solution, hyphenation.
Register variation is a cover term for “the full range of language varieties associated with
differences in communicative situation” (Biber & Finegan, 1993:316).
closeness to her by referring to their common heritage. This is despite the fact
that such a switch is less obvious to an observer than a change of language.
The associations of different varieties are sometimes consciously manipulated,
as in the case of advertisements, which often use CS into English to sell their
products (Chapter 4): see the German McDonald’s advertisement in Box 1.
The characteristic ways in which bilinguals combine their languages in a
particular community constitute a way of expressing their group identity – like a
characteristic accent. Both the languages themselves and the sociolinguistic
environment play a role in the patterns which emerge. Comparing CS across
different communities and different language combinations can help reveal the
relative role of linguistic and sociolinguistic factors – an important issue in
Linguistics. Within particular societies, sub-groups can be identified by their
characteristic CS patterns, as monolinguals can by discourse styles and registers. CS therefore helps us to understand identity formation and expression in
bilinguals (Tabouret-Keller, 1997; Sebba and Wootton, 1998).
Third, switching between languages provides crucial material for our understanding of how language is both comprehended (processed) in the brain, and
produced. What are the clues in the words and sentences we pronounce which
allow others to decode our meaning, and which we assemble in order to put
across that meaning? When we observe how this is done with two or more
languages, some of those features are thrown into sharper relief.
Fourth, by analysing code-switched speech, we can find out which combinations of words or morphemes from different languages can easily be combined and which are more resistant, or perhaps even impossible. Since grammar
consists of the rules regarding such combinations, CS acts as a signpost,
pointing at where the difficult issues may arise, and paving the way towards a
better understanding of grammar. Romaine, for example, has pointed out that
code-switching research can help us to understand a key issue in Linguistics:
the division of labour between grammar and lexicon (1995). Grammar specialists interested in CS try to discover whether the grammatical rules of the two
varieties in contact are sufficient to explain the patterns in mixed language
speech, or whether mixed codes have additional rules of their own.
All in all, CS is informative about language at a number of different levels.
There are also good reasons to study it in its own right.
Studying code-switching for its own sake
It seems sensible that linguists should derive their data and evidence from the
most typical speakers rather than from more exceptional ones. Numerous
linguists have pointed out that most of the world is plurilingual. If you add
together people who live in multilingual areas of the world (Africa, India,
Singapore, Creole-speaking areas such as the Caribbean or Papua New
Box 1 Code-switching in advertising
Advertisements are a very common locus for CS, and examples can be found
from round the world where English is combined with the local language in
order to evoke a cosmopolitan – or American – lifestyle. This advertisement
appeared as part of a code-switched series for McDonald’s in Germany in
the early 2000s. Note the use of many “bivalent” words (croissant, warm, so,
in), reinforcing the other similarities between German and English; the use
of German-style lower-case ‘a’ for the adjective american; and the use of the
masculine pronoun he, as in German, to designate the ham.
The lecker warm Croissant.
Geschnitten in two Teile, this is
very praktisch. So is genug
Platz for weitere leckere things.
The first leckere thing:
The Käse. A little bit
angeschmolzen and this
is very bequem for the
Schinken. He can not fall
out of the Croissant.
The second leckere thing:
The Schinken. Saftig and in praktische
stripes geschnitten.
McCroissant: the American answer to Croissant.
The delicious warm/warm Croissant.
Cut in/in two parts, this is very practical. So/so is enough room for further delicious
The first delicious thing: the cheese. A little bit melted and this is very comfortable for
the ham. He [sic] can not fall out of the Croissant.
The second delicious thing: the ham. Juicy and cut in/in practical stripes [sic].
Guinea, etc.); people who speak a regional language or dialect on top of a
national language (from Basques to Chechens); and migrants and their descendants (Greeks in Australia, Punjabis in Britain, Spanish speakers in the USA,
etc.), you are left with small islands of monolingualism in a multilingual sea.
This is without counting people who learn a second/third language beyond a
basic level at school (e.g. the Dutch or Scandinavians); those who have a
different language for literacy from the one they speak (e.g. Gujerati and
Punjabi speakers whose language of literacy is Hindi); those who become
bilingual through changes in personal circumstances; and those whose mother
tongue is not considered adequate for formal purposes (i.e. in diglossia) and
who therefore have to master another variety in order to take part in official life
(e.g. Flemish speakers in Belgium, speakers of dialectal Arabic in various Arab
countries) (Baker and Prys Jones, 1998). Plurilingualism is still the norm in
spite of the fact that a large number of the world’s languages are under imminent
threat of extinction owing to economic and globalizing forces (Crystal, 2000).
Most of these plurilingual speakers mix their languages in various ways in
their daily lives. CS has been studied from Mexico to Kenya and from Finland
to Australia – one of the main problems in writing this book has been the
difficulty of doing justice to the profusion of work which has been done. In line
with the eclectic approach to CS adopted here, a balance has been attempted
between giving up-to-date, accessible references and older/less accessible, but
seminal ones.
A common-sense approach
The approach to CS adopted here can be described as “common sense” or as
pragmatic with a small “p”. “Pragmatic” research on CS with a large “p” –
discussed in Chapter 4 – focuses on the conversational functions of CS and its
effects on conversational participants (Auer, 1998b). CS is presented here in a
rounded manner, looking at work carried out from the Sociolinguistic,
Psycholinguistic, Grammatical and Acquisitional perspectives as well as the
Pragmatic. CS is taken at face value, rather than with a particular theory as the
point of departure. It is important that CS be considered as the multifaceted
phenomenon it is, rather than purely as a means of testing theoretical positions.
So far, research on CS has been fragmented within various sub-disciplines,
yet there are considerable advantages in considering it as a whole. There is an
analogy with the poem at the beginning of this book, which recounts an old
anecdote about six blind men feeling different parts of an elephant, and being
unable to gain an overall view as to what it was. The terminological discussion
below (see 1.5) illustrates how little agreement there is about CS, its definition
and limits. Until greater concensus emerges, we should continue to look at it
from as many different angles as possible.
One reason for this is that each of the sub-disciplines in Linguistics uses
different methodological approaches. Sociolinguists seek to record “natural”
conversations but are subject to the Observer’s Paradox; theoretical linguists use
fallible intuitions as to which sentences are correct/acceptable; psycholinguists’
experiments test isolable – but incomplete – skills. None of these methods on its
own can provide a complete picture of behaviour as complex as CS.
Moreover, assumptions underlying tried and tested methodological paradigms are often insufficiently discussed in Linguistics. For example, variationist descriptions of CS are still current in the grammatical field (Poplack, 2000).
These are based on the assumption that if we can account for the majority of
cases of CS, then we are justified in ignoring the minority of cases which do
not fit in, and which are dismissed as aberrations. But one of the most famous
philosophers of science, Popper (1959), considered that a proposition which
was not falsified by counter-evidence was not scientific. Although there have
been a number of alternative proposals since then as to what constitutes proper
scientific enquiry, Popper’s test remains one of the most rigorous. At what
point do counter-examples to a paradigm which has been put forward make it
necessary to revise it? There has so far been a lack of discussion of this problem
with regard to CS, despite the fact that several scholars have argued that, in the
present state of knowledge, we should be trying to formulate grammatical
tendencies rather than absolute rules (Jacobson, 1998b; Muysken, 2000). As
Radford (1996:81) wrote: “Many advances in our understanding of adult syntax
have come from probing the syntax of structures … which any computer corpus
would show to be extremely ‘rare’ … ‘Every example counts!’” Tracy has also
pointed out, “What one counts as an exception is not just defined by some
quantitative feature but by the fact that it lies in conflict with our theory”
Another example from the grammatical field is the widespread belief that all
bilingual utterances have an underlying “Matrix Language”, i.e. a grammatical
template which can usually be identified with a particular language, such as
Russian, and that CS consists in grafting material from another identifiable
language, such as Spanish, onto such a base. This belief has been around at least
since Weinreich wrote that “Each utterance is in a definite language” (1953:7).
Such a belief makes the grammatical description of mixed language utterances
much simpler, but fails to address the more fundamental question of what we
mean by a language in the first place.
Le Page’s view is that one of the principal tasks for linguists is to explain the
formation of the concept of a homogeneous “language”: “We set out how we
saw such a concept evolving from observation of discourse, through the stereotypes denoted by such language-owning names as ‘English’ or ‘French’, to that
of the most highly abstract and focused Chomskyan ‘grammar’; and how actual
linguistic behaviour was influenced by the stereotypes as progressively it was
named, formalized, standardized, institutionalized, and totemized by a society”
(1997:31–32). “Languages” are often treated as if they were discrete, identifiable and internally consistent wholes, and we forget how historically recent and
culturally selective such a view is (see Chapter 2).
The study of CS should force us to think “outside the box”: to review
methodologies, theoretical approaches and assumptions, often developed in a
monolingual context, and see how they stand the test of being applied to the
speech of bilinguals. A common sense approach involves recognizing exactly
what are the limits to our ability to generalize at any given stage of the enquiry.
The study of code-switching
For a long time, CS was scarcely noticed by linguists writing about language
contact. Milroy and Muysken (1995), who describe it as “perhaps the central
issue in bilingualism research”, point out that research on CS was slow to start
compared with, say, research on borrowing or what used to be termed interference. In the seminal Languages in Contact (1953), Weinreich referred to the
“transfer of words” from one language to another by bilinguals, but dismissed
this as a “mere oversight” (1953:73–74). Haugen, writing at around the same
time, also apparently overlooked the significance of CS, and wrote that: “The
introduction of elements from one language into the other means merely an
alteration of the second language, not a mixture of the two” (1950:211).
Over the last forty-odd years, there has been an explosion of interest in CS.
CS had remained more or less “invisible” in research on bilingualism until the
work of Gumperz and his associates in the 1960s and early 1970s (Gumperz,
1964, 1967; Gumperz and Wilson, 1971; Blom and Gumperz, 1972). Thereafter
the subject took off – and there has been no sign of a downturn – as people realized
that CS was not an isolated, quirky phenomenon but a widespread way of speaking. But research in this field is complicated by the multilayered significance of
CS. Each new case which is documented can be looked at from multiple perspectives, so from the outset, a certain depth of engagement with the data is necessary.
Furthermore, by definition, studying CS implies dealing with several languages.
Grasping the significance of a transcription where the reader or researcher is
not familiar with one or both of the languages involved can be off-putting. This
problem should be somewhat reduced in the future by various technical developments of use to the linguistic researcher, such as standardized transcription and
coding systems, sound–text linking, and the possibility of collaborating on and
sharing data over the Internet. Proposals for a system appropriate for CS are
summarized in the Appendix (LIPPS Group, 2000; Gardner-Chloros, Moyer
and Sebba, 2007). Because of the huge interest in CS on the one hand, and the
difficulties of studying it on the other, a lot of work has crystallized around a few
main approaches:
(1) Sociolinguistic/ethnographic descriptions of CS situations. These represent
the majority of studies of CS. Although by their nature, they remain fairly
fragmented, many important insights are derived from linking the manifestations of CS to aspects of the sociolinguistic situation (Chapter 3).
(2) Pragmatic/conversation analytic approaches. These rely on identifying
the meanings brought about by CS in conversations, for example through
following, or avoiding, the language choices of interlocutors (preference
organization). This use of CS complements the exploitation of contrasting
connotations of the two varieties (e.g. we-code/they-code). Such tactics may
be used in the same conversation (e.g. Milroy and Gordon, 2003; McCormick,
2002) (Chapter 4).
(3) Grammatical analyses of samples of CS and the search for underlying
rules, models and explanations to explain the patterns found. These
have developed largely as a separate tradition from the sociolinguistic
and the pragmatic. Although some authors have identified connections
which deserve to be investigated (Muysken, 2000; Myers-Scotton,
1993b), this has not been a primary focus in the research so far
(Chapter 5).
Each of these approaches is the subject of one chapter, as is the place of CS in
language contact (Chapter 2); the implications of psycholinguistic work on
bilinguals for our understanding of CS (Chapter 6); and CS in children and
other learners (Chapter 7). The chapter pattern reflects the main research output
and traditions, but within each chapter it is emphasized that there are no strict
divisions between the questions which should be addressed in CS research – on
the contrary, a major purpose of this book is to encourage the formulation of
more holistic insights and research.
The vexed question of terminology
In the introduction to a volume on CS, Eastman wrote: “Efforts to distinguish
code-switching, code-mixing and borrowing are doomed” (1992:1). Little has
occurred since then to lighten this pessimistic view: terminology has been
endlessly discussed in the CS literature without any real commonality of
practice being achieved. Several good descriptions are, however, available
of how the most important terms have been used in this field (Milroy and
Muysken, 1995; Li Wei, 2000; Hamers and Blanc, 2000; Clyne, 2003), the
key issues being described in sections 1.5.1 and 1.5.2.
A misleading term?
CS is not an entity which exists out there in the objective world, but a construct
which linguists have developed to help them describe their data. It is therefore
pointless to argue about what CS is, because, to paraphrase Humpty Dumpty,
the word CS can mean whatever we want it to mean.
Janički (1990) drew attention to the dangers of essentialism in sociolinguistics, illustrated by the use of “left to right” rather than “right to left” definitions.
A “left to right” definition states that such and such a word (which is, as it
were, on the left-hand side of the page as you write) is this, that or the other, for
instance that CS is the alternate use of two languages in conversation – as if this
were an objective or immutable truth. Instead, he claims, sociolinguists should
make use of “right to left” definitions, e.g. “We will call the alternation of
two languages in conversation code-switching”, where the term which we are
defining is on the right-hand side. This makes the definition into a working
definition rather than suggesting that we are imparting an essential truth. This
advice is helpful given the confusing fact that each researcher provides different
definitions of the relevant terms. This matters less if we consider definitions
merely as a research tool with which to describe data.
Ideally, we would of course like to move forward, as researchers, speaking
the same theoretical language. It would be preferable if the most salient phenomena which we observe had a name, which was not seriously misleading
as to their nature. Unfortunately, both halves of the term CS are misleading.
“Code” was originally taken from the field of communication technology (Fries
and Pike, 1949; Fano, 1950; Jakobson, Fant and Halle, 1952). What was meant
there has nothing to do with language: “code-switching” refers instead to a
mechanism for the unambiguous transduction of signals between systems (for
the history of the term, see Alvarez-Cáccamo, 1998; Benson, 2001; TabouretKeller, 1995). Nowadays code is understood as a neutral umbrella term for
languages, dialects, styles/registers, etc., and partly usurps the place of the more
usual “catch-all” term variety to cover the different sub-divisions of “language”.
Benson describes work on CS as we understand it now, carried out in the early
years of the twentieth century, as “forgotten”, because scholars have been
insufficiently interested in the history of their field.
“Switching” appears transparent enough, in that it refers to alternation
between the different varieties which people speak.5 In the early psycholinguistic studies of bilingualism in the 1950s and 1960s, psychologists assumed
that something similar to flicking an electric switch went on when bilinguals
changed languages. More sophisticated models followed, in which, for example, a different switch mechanism was said to control input (listening and
understanding) from that which controlled output (speaking) (see Chapter 6).
But accumulated evidence from the mixed speech of bilinguals has led to the
transition between the two varieties being seen as more and more complex and
In French the term alternance codique is used, in German Kodewechseln, in Dutch Kodewesseling –
however in all these, the English code-switching is often used instead.
less and less clear-cut. Clear-cut changes, such as the term “switching” implies,
may occur in bilingual and bidialectal speech but they are only one of many
possibilities (see Chapter 6).
What does code-switching cover?
Milroy and Muysken (1995) wrote in the introduction to One Speaker, Two
Languages, that, “The field of CS research is replete with a confusing range of
terms descriptive of various aspects of the phenomenon. Sometimes the referential scope of a set of these terms overlaps and sometimes particular terms
are used in different ways by different writers” (p. 12). The same problem was
outlined in Clyne (1987:740–741). He pinpointed the crucial distinction
between those who consider CS to be “fuzzy-edged”, i.e. on a continuum
with respect to borrowing, syntactic merging, etc., and those who consider it
as the one form of language contact which does not involve convergence of
the two systems. More recently, Clyne has suggested that we should reserve
code-switching for transference of individual lexical items through to whole
stretches of speech, but that we should adopt a different term – he suggests
“transversion” – for cases where the speaker “crosses over” completely into the
other language (2003:75). In his survey of grammatical studies of CS, Muysken
(2000) reserves the term CS for one of the three types of language mixture
which he describes: alternation. He uses code-mixing for the other types,
insertion and congruent lexicalization (see Chapter 5).
Haugen (1956), one of the first to write about CS, distinguished between
code-switching on the one hand, in which the character of the contributing
varieties is said to be preserved, and interference and integration on the other.
This distinction was in large measure taken over by Poplack and Sankoff (1984)
and Poplack (1988). They, however, used the term borrowing or nonce borrowing for instances of juxtaposition which do show some measure of convergence,
be it morphological, phonological or syntactic. Others have argued that there is
no clear line between CS and borrowing (Gardner-Chloros, 1987; MyersScotton, 1992; Thomason, 2001; Treffers-Daller, 1994) and that the two are
on a diachronic continuum: loans start off as code-switches and then gradually
become established as loans (see further discussion in Chapter 2). Yet another
point of view is that speakers need to be evaluated individually in order to
decide whether a particular morpheme is a one-off occurrence, a conscious
code-switch or a loan. “What appears to be a nonce borrowing, or an occasional
code-switch, for one speaker, could be an established morpheme for another
speaker” (Aikhenvald, 2002:197).
A frequently made distinction is between code-switching and code-mixing,
though here too, the line has been drawn in different ways. Some have reserved
code-switching for cases where the two codes maintain their monolingual
characteristics and used code-mixing for those where there is some convergence
between the two (e.g. Muysken, 2000). Confusion arises, however, because the
two processes often co-exist within the same stretch of discourse as well as
overlapping at the conceptual level (Hamers and Blanc, 2000). Sridhar and
Sridhar (1980) and Bokamba (1988) use code-mixing for alternation within the
sentence and code-switching for alternations going beyond the sentence borders. Meisel (1989) employs code-mixing for the fusion of two grammatical
systems, whereas he describes code-switching as the pragmatic skill of selecting
the language according to the interlocutor, topic, context, etc.
Further variations can be found. In social psychology, CS refers to language
choices in a bilingual setting (Sachdev and Bourhis, 1990). Meeuwis and
Blommaert (1998) point out that CS itself can be a variety on its own, with the
same functions and effects as those usually attributed to “languages”. Finally, as
here, CS may be used as a general term covering all outcomes of contact between
two varieties, whether or not there is evidence of convergence. Such an approach
is justified in that convergence can occur at many different levels and it is not
in practice always possible to decide where it occurred and where it has not
(Neufeld, 1976; Pfaff, 1979).
Also, of the various terms which have been used, CS is the one which has
gained the widest currency. This has led to a situation where the study of CS
is sometimes a victim of the success of the term CS. The term encourages
people to think of language contact in terms of discrete alternation between
two self-contained systems, and to neglect the connections between that
type of occurrence and others where convergence occurs. Examples of the
latter include loanwords, pidgins, mixed languages, etc., and the concept is
crucial to understanding the historical development of languages, e.g. English
as the product of Anglo-Saxon, Norman, Latin and other influences. The
“new orthodoxy” of believing that CS is made up of alternation between
discrete systems has, for some researchers, taken over from the “old orthodoxy” of ideal speaker/hearers in homogeneous communities (GardnerChloros, 1995).
Studies of code-switching
Substantial chapters or sections have been devoted to CS in the principal
volumes on bilingualism and language contact, e.g. Romaine’s Bilingualism
(1995), Coulmas’s Handbook of Sociolinguistics (1997), Hamers and Blanc’s
Bilinguality and Bilingualism (2000), Thomason (2001), Clyne (2003) and Li
Wei (2000). Various edited collections and special issues of journals have been
devoted to different aspects of CS (Heller, 1988a; Eastman, 1992; Milroy and
Muysken, 1995; Auer, 1998b; Jacobson, 1998a, 2001; Dolitsky, 2000, Li Wei,
2005), as well as a large number of academic papers. A search of titles in the
Language and Linguistic Behaviour Abstracts on the term code-switching just
for the last five years produces several hundred titles. To this, we can add several
full-length monographic case-studies of bilingual situations where CS is the
main focus (Agnihotri, 1987; Backus, 1992; Bentahila, 1983; Gardner-Chloros,
1991; Gibbons, 1987; Halmari, 1997; Haust, 1995; Heath, 1989; Myers-Scotton,
1993a; Nivens, 2002; Nortier, 1990; Treffers-Daller, 1994; McCormick, 2002;
Zentella, 1997), covering a variety of language combinations. Two books
have been devoted to the grammatical aspects of CS (Myers-Scotton, 1993b;
Muysken, 2000), as well as a further volume by Myers-Scotton to developing her
grammatical theory in the broader context of language contact (2002a). CS is
now a prolific area of research, and the subject of many doctoral and master’s
theses worldwide. It figures prominently in the principal journals on bilingualism, notably the International Journal of Bilingualism and Bilingualism,
Language and Cognition.
With a few exceptions, notably the European Science Foundation Network
on Code-switching and Language Contact (1988–91), which produced three
volumes of papers,6 and the biennial series of Symposia on Bilingualism started
by Li Wei in 1999, the field has, in general, been marked by a lack of coordination and dialogue between researchers. A project is now underway to set up a
database of CS texts for researchers to share and to provide material for comparisons (LIPPS Group, 2000). As was pointed out in a paper describing the aims of
this project, “The great majority of … studies involve the collection of new sets
of data by individual researchers; so while there is a lot of often painstakingly
collected data around, it seems likely that the endless collection of more data for
analysis will no longer constitute the most productive application of research
efforts” (Gardner-Chloros, Moyer, Sebba and van Hout, 1999).
Speakers’ insights
Ethnographic research on CS involving field-work in various communities and
the gathering of different types of data – attitude studies and interviews as well
as recordings – provides various insights into CS from the speakers themselves.
Some of these insights merit particular attention:
(1) Bilingual subjects sometimes spontaneously provide an explanation as
to why CS occurs: laziness. This explanation is not only offered by
speakers who claim not to switch themselves, but also by self-confessed
code-switchers. CS is seen as an easy way out when people cannot be
bothered to search for the words they need in a single language. As
laziness is not generally considered a virtue, this interpretation obviously
The volumes themselves are no longer available, but led to the publication of Milroy and Muysken
depends on believing that it is in some way “wrong” to mix languages.
Irrespective of such normative attitudes, it is of interest, from a psycholinguistic and from other perspectives, to consider whether code-switching
involves less or more effort than monolingual speech.
(2) Regular code-switchers often claim to disapprove of CS when specifically
asked. Surprisingly few systematic studies of attitudes to CS have been
carried out (see Chapter 3), but where they have, they have generally
confirmed that people are not proud of it. Approval of CS tends to coincide
with a laid-back attitude towards authority; for example within the same
community it tends to be more common in younger rather than older
generations (McCormick, 2002; Gardner-Chloros, McEntee-Atalianis
and Finnis, 2005). But contrary to some popular belief, it is not certain
that CS is used more by uneducated people than by educated ones – in
several cases, the opposite has been shown (Timm, 1978; Haust, 1995).
(3) Although people who live in bilingual communities are, generally, aware of
the existence of CS and of the fact that they themselves sometimes switch,
their level of awareness of their own CS behaviour seems to lag far behind
their practice. When recordings of code-switched conversations are played
to the subjects involved, for example so as to enlist their help with the
transcription, they frequently express surprise and/or embarrassment on
discovering the extent of their own mixing.
To sum up, CS is:
(1) thought to be an easy or lazy option;
(2) generally disapproved of, even by those who practise it;
(3) below the full consciousness of those who use it.
Such insights from speakers can be useful for formulating research hypotheses. For example, let us look briefly at the first point, whether CS is really the
easiest option. Sociolinguistic studies do show that people code-switch more,
and more within the clause, when they are at ease, in informal situations
(Gardner-Chloros, 1991:186; see also Chapter 3). Dewaele (2001) also found
that second language learners code-switch more in informal than in formal
interviews. But does this mean it is actually easier to switch than not to switch?
The answer depends on a variety of factors. Where the speaker is a balanced
bilingual – i.e. someone who can speak either variety equally competently, CS
may be used deliberately as a compromise strategy, when addressing others of
varying competences and preferences. It might not then be the easiest solution,
but merely the most expedient. At a psycholinguistic level, the position is also
less than clear. Inhibiting one of the languages – i.e. preventing it from coming to
the surface – does apparently require some kind of effort (or “resource”), but the
joint activation of two varieties does so as well (Green, 1986/2000). The mental
effort required for the simultaneous, or rapid successive activation of two
competing systems translates into extra split seconds of time which are required
in both the receptive and productive modes (see Chapter 6). The jury must
therefore remain out on whether CS is always the solution requiring least effort.
For the speaker in Example 2 above, it is clear that the least effort would have
involved speaking monolingual English.
What the speakers’ own views about CS do point to, is a dissociation between
how they use their linguistic competence (which is dictated by expediency,
pragmatic considerations, habit and a variety of other sub-conscious motivations),
and what they know or think they know about it (which includes their overt
attitudes, purism, etc., as well as their representation of how they speak). The
underlying competence of the bilingual is a poly-idiolectal personal amalgam,
influenced only indirectly by linguistic attitudes, which are usually shared with
others in the community. This “multilayered” aspect of language within the
individual is consonant with Le Page’s remarks regarding the different senses
of a “language” (see above).
Further types of code-switching
Until recently, descriptions of bilingual speech and, more generally, much of
the research on bilingualism tended to mention only in passing and in a fairly
speculative manner how findings might differ in the case of trilinguals, and
speakers of more than three languages. Instead of bilingualism being seen as
one possible case of plurilingualism or multilingualism (which, in theory,
appears logical), the term “bilinguals” is often used to subsume “plurilinguals” (e.g. Sachdev and Bourhis, 2001:407). Speakers of three or more
languages are by no means rare in the world, still less so if one considers
competence in several varieties such as different dialects as well as those
who speak distinct “languages”. They have only recently begun to be considered more systematically (Hoffmann, 2001; Cenoz, Hufeisen and Jessner,
Example 4
I have canné todo
failed everything
(Baetens Beardsmore and Anselmi, 1991: 416)
This example involves quadrilingual components: pronoun subject and auxiliary in English (I have), main verb stem in Italian (cannare ‘to fail’), French
verb ending (-é for past participle) and pronoun direct object in Spanish (todo
‘everything’). Lest this be thought exceptional, Clyne (2003) shows how, on the
contrary, the fact of switching once actually creates the possibility of further
switching: instead of going back to the variety used before the switch, trilingual
speakers often take a different “branch” on “exiting” from it and switch to a
third language, as in Example 5.
Example 5
[G = German, E = English, D = Dutch]
Ich muss ab und zu in einem [G] dictionary [E] KIJKEN [D]
I have to every now and then in a dictionary look+INF
‘I have to look in the dictionary every now and then’
(Clyne 2003:163)
Clyne suggests that it is because the word dictionary is a lexical transfer in the
speaker’s German as well as in their Dutch that a “transversion” to Dutch is
facilitated for the main verb, kijken ‘to look’.
Baetens Beardsmore has further shown that, in complex multilingual societies such as SE Asia or Africa, relative dominance in the different languages
in a speaker’s repertoire may not coincide with chronological order of learning. A second language may be the source of interference or CS into the third,
the first learned language being restricted to intimate family life. Although
determinants of transfer are the same however many languages are involved,
“an interweaving of linguistic and extra-linguistic variables decide which of
the languages serves as the source of the transfer” (1986:82). These examples
provide a small idea of the many further questions – mainly unexplored –
which arise in relation to CS with more than two varieties.
Second language learners
While there is a considerable body of research about the learning processes of
second language (L2) learners (Mitchell and Miles, 2004), as yet, relatively
little has been written about their CS, discussed in Chapter 7 (Poulisse and
Bongaerts, 1994; Dewaele, 2001). Since even “native” bilinguals are rarely
balanced in their competence in the two varieties, there seems little reason in
principle to draw a clear line between them and L2 learners, and it is to be
hoped that, in the future, further systematic comparisons will be made of
learners’ CS with that of “native” bilinguals. Late acquisition of the L2 appears
to make less difference to a speaker’s CS than the extent to which they belong to a
code-switching community. This is demonstrated in the case of members of
immigrant groups, whose CS can be just as intense as that of native plurilinguals.
There remains much to be learned about CS involving different types of
plurilingual speaker, and there may be many insights to come from such studies
which will affect our understanding of CS as a whole. The same applies to CS
between different modalities (e.g. between oral and sign languages) and doubtless to switching between other, non-linguistic forms of behaviour (Chapter 8).
CS is a growth area in linguistics, since it provides insights not only about
plurilingualism – which concerns the majority of speakers in the world – but
also about language itself, from several different perspectives. This has yet to be
recognised by some linguists of a theoretical persuasion – or monolinguistic
orientation – who treat the questions posed by plurilingualism as marginal to
linguistics. In The Twitter Machine, Smith remarks: “It is obvious that different
communities exhibit variation in their speech: people in Paris speak French
while those in Washington speak English and those in Montreal cope with both;
it is equally clear that children don’t speak the same way as their grandparents,
that males and females are not necessarily identical in their linguistic abilities,
and so on. In short, any social parameter whatsoever may be the locus of some
linguistic difference. Unfortunately, nothing of interest to linguistic theory
follows from this” (1989:180).
It is nothing new for theoretical linguists to express lack of interest in
performance. But of all aspects of performance, CS may be the one where the
notion of “acceptability” is hardest to maintain, as it challenges the whole
notion of the “native speaker”. Code-switchers upset the notion of performance
errors by contravening and rewriting the expected rules. The ideal speakerlistener is an elusive figure indeed in many bilingual communities.
Linguists who do study bilingual performance, in a sociolinguistic context
for example, have shown how CS contributes in various ways to an understanding of how the individual is articulated with the social. They too have to be
aware of the limitations imposed by continuing to use well-worn methodological approaches and failing to consider whether such approaches address the
right questions. Studies of language and gender provide a good example of the
dangers inherent in doing this – it is thanks to the imaginative use of different
methodologies that it is now clear that gender as such is only an intervening
variable, and not an explanation in itself, for many differences in performance
between women and men (Coates, 1993; Eckert, 1989).
Conversation analysts specify that they only consider as CS those transitions
in one speaker’s discourse which can be shown to have some identifiable impact
on the conversation. Researchers looking at CS from other perspectives have
not, on the whole, specified so clearly their method for pinpointing the object
of study. Is it sufficient for an observer to identify two different entities, or must
the distinction be based on insider – even idiolectal – knowledge? CS research
has not solved these problems, but it does confront us with them very starkly.
If we can find ways of tackling these issues, this will be useful for many other
aspects of linguistic research as well as for CS.
As we have seen, a central question is what implicit definition of “language”
we are applying. This goes beyond the question of the best methods of analysis
for analysing bilingual speech and raises a philosophical problem. The notions
of focusing and diffusion, developed by Le Page (1989; 1997) and Le Page
and Tabouret-Keller (1985), drew inspiration from the creole-speaking areas
where much of their research was carried out, but were intended to have a
general application. CS gives us both material and incentive to carry forward Le
Page’s challenge to settings where “diffuseness” still operates, if less obviously
than in creole areas.
If we can rise to both sorts of challenges discussed here, the methodological
and the ideological, then enhancing our understanding of CS will have a variety
of repercussions in linguistics. As we will see, so far such breakthroughs have
come more from grass-roots descriptions of actual data than from the top-down
application of theoretical frameworks developed in more idealized settings. It
is for this reason that a “common sense” approach is advocated.
Code-switching and language contact
This chapter describes the place of CS in language contact, including its relationship with borrowing, pidginization, convergence and language shift. All of these
are instances of change, but the time-scale in which such change occurs can vary
widely. CS occurs in contact situations of many types and relates in complex
ways to the processes of change at work in those situations.
CS occurs among immigrant communities, regional minorities and native multilingual groups alike. Gumperz and Hernandez wrote that it could be found “each
time minority language groups come into contact with majority language groups
under conditions of rapid social change” (1969:2). Others (e.g. Giacolone Ramat
1995) have on the contrary described it as a feature of stable bilingualism in
communities where most speakers can speak both languages. There are descriptions of contact situations in which it receives little or no prominence (e.g. Jones,
1998, with respect to Wales; Spolsky and Cooper, 1991, on Jerusalem), but this
does not necessarily mean it does not occur in those settings. Instead it may be
because other instances of contact or restructuring are the primary focus, or
because the data collection techniques do not centre on the informal conversational modes where CS occurs. In Section 2.2, the extent to which CS brings
about language change, or is symptomatic or independent of it, is considered.
Sociolinguists have treated CS mainly as a spoken genre, and as such it
undoubtedly has a long history (see Stolt (1964) on German–Latin CS in
Luther’s “table-talk”). But CS is also found in written texts from various historical
periods (Montes-Alcala, 1998; Sebba, 2005). Examples include Latin–Greek
in Cicero’s letters to his friend Atticus (see Box 5), French–Italian in a thirteenthcentury Coptic phrasebook and a fourteenth-century Venetian manuscript of the
Song of Roland (Aslanov 2000); German–Latin in the work of a seventeenthcentury linguist, Schottelius (McLelland, 2004); English–French CS in a variety
of Medieval English texts (Trotter, 2002) (see Box 3); through to literary works
such as Chicano poetry (Valdes-Fallis 1977) and novels where spoken CS is
represented, such as Tolstoy’s War and Peace (see Timm, 1978), Eco’s The Name
of the Rose (where Salvatore speaks a multilingual jargon), and through to
contemporary novelists such as Zadie Smith (English–Creole in White Teeth)
Code-switching and language contact
and John Markovitch (Spanish–Quechua in The Dancer Upstairs). Scriptswitching is also found, as when Jan van Eyck wrote ‘Als ich kann’ in German
(‘as (well as) I can’) above his self-portrait in a turban, but in Greek letters: ‘aλς ιχ
χάν’, perhaps to indicate his erudition.1 Angermeyer (2005) has described script
alternation in Russian–American advertising, as a way of signalling a bilingual
identity. A Hassidic newsletter distributed in North London, Hakohol, includes
code-switches from Yiddish and Hebrew, the latter in Hebrew script (see Box 6).
The lifestyle magazine Latina for American women of Hispanic descent is full
of Spanish–English CS, which Mahootian (2005) describes as a reflection of
community norms, but also as challenging relations of power and dominance
between the older Hispanics, the monolingual English-speaking community and
the younger Hispanic generation (see Box 2). It is also symptomatic of certain
types of written discourse taking on the informality of conversation, as in the email
and texting practices of many young people of mixed heritage (Hinrichs, 2006).
There is even a “mini-genre” of “bivalent” texts, dating from the sixteenth
to eighteenth centuries, composed so that they could be read simultaneously in
Latin and various Romance languages, notably Spanish – a way of “punning” with
linguistic affiliation rather than with meaning (Woolard and Genovese, 2007).
Code-switching as a symptom of different/opposite
CS arises in a variety of different contexts, as a symptom of quite opposite
developments, from accommodation to divergence and from language maintenance to language shift. It reflects social differences and tendencies within the
same society and language combination (Bentahila and Davies, 1991; Li Wei,
1998a; Treffers-Daller, 1998), just as it reflects those between different societies
and different language combinations (Poplack, 1988; McClure, 1998). Ideally,
CS should be seen as part of a “bigger picture” including other forms of register
variation (Halmari and Smith, 1994).
Convergence v. preserving distinctiveness
An important question regarding the place of CS in language contact is to what
extent it is a mechanism for bringing the varieties closer together. As we saw
in Chapter 1, it has sometimes been defined – misleadingly – as the one type of
language interaction where each variety preserves its character. At the same
time, data collected in many communities has shown that there is no one-toone correlation between CS and language change or shift. Thomason (2001)
‘ιχ’ is also a pun on his name (Eyck).
Box 2 Code-switching in the ‘ethnic’ press
Publications intended for communities of immigrant origin are often rife with
CS. The magazine Latina, addressed to young Hispanic women in the USA,
started in 1996 and has a circulation of 175,000. Mahootian (2005) claims that
in this context, CS is used “as a direct and undeniable assertion of bilingual
identity” and that “linguistic necessity is not the driving force” behind it –
rather it is evidence of its acceptance and propagation, i.e. of linguistic change.
Article taken from Latina magazine, which is aimed at young Hispanic women in the
USA, March 1999
Code-switching and language contact
gorditas: lit. ‘little fat ones’, here used as a term of endearment to refer to chunky/chubby
flaca(s): skinny(ies)
mujeres con curvas: women with curves
mujerona: big woman
amazona: amazon
y Ave María: and hail Mary
viva gorditas: (long) live chubby women
This article is discussed in Mahootian (2005).
has considered this question at length and concludes that CS is not a major
mechanism in contact-induced change, as it does not result from “imperfect
learning”, though it is one of the principal mechanisms of borrowing. She is
at pains to point out how varied and heterogeneous the sources of language
change are, and that most of the general rules are there to be broken. For
example, language change often takes place when a minority group becomes
bilingual and adopts features of the L2, as in the case of Greek spoken in
Asia Minor adopting features of Turkish. As a general rule, the language of
the majority is adopted by the minority rather than vice-versa. However, this
tendency is in conflict with another one, i.e. that the language of the elite is
adopted by the subordinate group, and the elite is, almost by definition, the
minority. Thomason points out that native speakers of Turkish may also
have played a role in bringing features of Turkish into Asia Minor Greek,
as some had learned Greek as a second language, in spite of being the
majority (2001:67).
Backus (2005) provides a description of the complexity of the relationship between CS and structural change. He lists numerous problems, both
methodological and conceptual, with the hypothesis that structural change
is directly brought about by CS. For example, at a methodological level, one
would need to have a complete picture of the range of variation of the structure
within the monolingual variety in order to establish that it was indeed CS, and
only CS, which had brought about the change. At a conceptual level, one
would need to understand how exactly the fact of code-switching could bring
about a change in the structure of a language, and distinguish this from calques
(i.e. expressions taken over literally, either word for word or morpheme for
morpheme, from one language to another) in the idiolect of the individual
Studies which show code-switching to be bound up
with shift or change
In most communities where there is CS, there is a correlation between the speakers’ age and the type of CS which they use. For example Bentahila and Davies
(1991; 1998), in analysing the CS of Arabic–French bilinguals in Morocco, found
clear differences in the patterns characterizing the younger and the older groups,
related to differences in the role played by French in their education and background. In a study of Arvanitika, a dying Albanian dialect spoken in Greece,
Trudgill (1977) found that CS into Greek made it very difficult to identify which
aspects of Arvanitika were being lost because CS was being used as a compensatory strategy. Lavandera (1978) reports a similar phenomenon in Buenos Aires,
where migrants of Italian origin switch between their own dialect of Argentinian
Spanish, Cocoliche, and Standard Argentinian Spanish to compensate for
reduced stylistic options available to them in either variety. Schmidt (1985)
shows how among the older, more fluent speakers of Dyirbal, a dying
Aboriginal language of Australia, only individual words are said in English,
e.g. all right, now, finish (‘period’). Moderately proficient speakers use
many more imported English words in their Dyirbal, which are often adapted
to Djirbal morphology, e.g. ring-iman ‘she phoned him’, jayil-gu ‘to jail’, one
night, down there, etc. Finally the younger speakers, who are much less fluent
in Dyirbal, code-switch copiously between the two varieties.
Example 1
We tryin’ to warn
ban wuigi nomo wurrbay-gu.
We were trying to warn her not to speak [Dyirbal]
(Schmidt, 1985)
Apart from intergenerational comparisons, there is also evidence from historical
linguistic work which suggests that CS is an important component of change.
This has been notably been studied in the context of Middle English, which
relexified2 under the influence of Norman French and Latin. “Bilingualism and
CS must have played a major role in the process of lexical borrowing and
mixed-language texts can thus provide interesting information on the process of
widespread relexification of English in the ME period. As Rothwell pointed out,
“Generations of educated Englishmen passed daily from English into French
and back again in the course of their work”, a process which must have led to
specific lexical transfers both in the field of technical and of general vocabulary” (Schendl, 2002:86). Schendl’s paper gives numerous and varied examples
of CS between English, French and Latin from a variety of medieval genres
(business, religious, legal and scientific texts) to back up this claim.
Relexification is the process by which a language which retains its basic grammatical identity
replaces part or all of its lexical stock with words from another variety.
Code-switching and language contact
It is important to point out that CS takes place in a context where there is shift in
progress, rather than constituting shift of itself. This was shown in Gumperz’s
early study of Hindi–Punjabi CS in Delhi (1964), where two related varieties
had substantially converged at a grammatical level, through being in close
contact for decades, while retaining a different lexical stock. As a number of
specific differences between them remained, Gumperz concluded that these
differences must be functional, as they would long ago have been wiped out if
they were not serving a specific purpose.
In Rindler-Schjerve’s (1998) study of Italian–Sardinian CS, she emphasizes
that although the switching occurs in a context of language shift, it “should not
be seen as a mechanism which accelerates the shift” (1998:247).3 In Sardinia, it
is the more balanced bilinguals who switch most, and who “contribute to the
maintenance of Sardinian in that they change the Sardinian language by adapting it to the majority language thus narrowing the gap between the two closely
related codes” (1998:246). In Example 2, the Italian expression secondo me ‘in
my opinion’ is inserted in a Sardinian sentence, but adapted to Sardinian
phonology to minimize the transition (segunnu me). Its function is to highlight
or separate the parenthetical expression ‘in my opinion’.
Example 2
Non m’an giamadu ‘e veterinariu ma segunnu me fi calchicosa chi a mangiadu
They didn’t call a vet but
in my opinion it was something
which it has eaten
(Rindler-Schervje, 1998:243)
A sign of doom or of vitality?
CS can arise in situations of widely varying stability. It can be a feature of stable
bilingualism for an extended period, and then, following social changes, it may
persist and become implicated in language shift. The shifting and the shifted-to
varieties are necessarily in contact, at least in some sections of the community
and/or contexts, over a period of time: This is the case of Sauris in the Carnian
Alps, described by Denison (1984; 1986). This small community had preserved
a local variety of German for some 600 years, Friulian being the regional norm
and Italian the national norm, with which speakers, until this century, had
relatively little contact. Now the local dialect is declining at the expense of,
above all, Italian, but is as yet evident in the frequently code-switched speech of
the community. “Functional and substantial linguistic substitution within a total
repertoire can (and usually does) proceed selectively, over quite a long period,
My italics.
involving many generations, before the stage is reached when what was once an
entire linguistic tradition … is … jettisoned” (Denison, 1987:73).
Change can be fast or slow and can affect all aspects of language (Gumperz
and Wilson, 1971; Dressler and Wodak-Leodolter, 1977). It can take place over
several generations or, effectively, within a single generation. It may be difficult
to detect, losses being masked by code-switches (Trudgill, 1977) or internal
restructuring (Tsitsipis, 1998), and it can occur either with heavy linguistic
symptoms such as morphological loss or without them (Dorian, 1981; Schmidt,
1985). A range of linguistic configurations can arise along the road to extinction. There are descriptions of the characteristics of semi-speakers’ competence,
such as loss of subordinative mechanisms, word-retrieval problems and phonological distortions or hypercorrections in relation to a number of languages and
sociolinguistic settings (Tsitsipis, 1998; Schmidt, 1985). Dorian (1981), however,
points out that there are circumstances where, depending on the speed with which
a variety is abandoned, languages may die without their morphology having
altered at all, and Tsitsipis argues against any “direct correlation of structural
intactness with the functional survival of a language” (1998:65).
Auer (1999) describes CS as the first point in a chronological progression
along a continuum. At the CS stage, the point in the sentence where there is
a switch is a significant aspect of the conversation. The next stage is language
mixing, where, as in Example 2 above, it is not the individual switch
points which carry significance, but the use of the overall switching mode –
this stage is also described by Myers-Scotton (1993a) as “switching as an
unmarked choice”. The third stage is fused lects, which are stabilized mixed
In conversation analysis, the “proof” that speakers are drawing on two
separate systems lies in the reactions of their interlocutors to the contrasts and
conversational effects which they bring about. Thus Auer’s definition of CS is
“the juxtaposition of two languages perceived and interpreted as locally meaningful by participants” – here “locally meaningful” applies to the juxtaposition
itself. In language mixing on the other hand, the combination of varieties does
not give rise to local, but to more global, meanings. Individual switches no longer
serve a conversational purpose as in Example 2 above, but the distinction
between the languages has not collapsed completely. In the third stage, the
development of fused lects, there is a loss of variation: the use of elements
from one or other variety is no longer a matter of choice, but of grammatical
convention. Structures from each variety, which are equivalent in monolingual
usage, develop specialised uses.
Auer claims this process is unidirectional. It may never be completed, as
bilingual communities may stabilize at any point along the way, but it does not
allow for any movement in the opposite direction. Auer goes so far as to say that
the movement from fused lects to language mixing to CS is “prohibited”.
Code-switching and language contact
If you begin with a fused lect – i.e. a fully conventionalised mixed variety,
presumably with no more internal variation than any other natural language –
it is indeed difficult to see how this could spontaneously “split up” into its
historical component parts. The social circumstances which gave rise to the mixing
cannot be “put into reverse”: English is not likely to split up into Anglo-Saxon,
Norman French, etc. On the other hand, if the starting point is two varieties which
are somewhere towards the middle of the continuum, say between CS and
language mixing, then these could in theory split or become more similar to
(one of) the component varieties; this is what happens in decreolization. If we
take the example of the Creole–London English mixture described in Sebba
(1993), one of the elements, London English, has an independent existence anyway, and it is conceivable that this type of creole could begin to be used
independently, if the identity associated with it were to develop further.
We should be careful to distinguish situations where there is a progression
from one state to another from those where all we know is that different
phenomena overlap chronologically. We will see in Chapter 3 that different
kinds of CS can co-exist in the same community at the same time. As Crowley
pointed out with respect to pidginization in Bislama (1990:385): “Rather than
trying to divide the history of Melanesian Pidgin into chronologically distinct
developmental phases, we should regard stabilization, destabilization, and
creolization as all contributing simultaneously to the gradual evolution of the
language from the very beginning.”
There are also circumstances where contact-induced change does not proceed
smoothly through all three stages. After some convergence has taken place,
instead of the old variety being abandoned in favour of the new, the altered
(code-switched) variety, brought about through contact, may assume distinct
functions of its own. This type of case was discussed by Gumperz (1964) in
relation to varieties in contact in Delhi, and may explain the formation of certain
The stabilization of CS varieties arises when these varieties assume an
identity function. They are often characteristic of young second-generation
immigrant communities which develop a pride in their mixed identity. Such an
intra-community variety is, for example, emerging among the second-generation
Portuguese in France. Known as “immigrais” (immigrese), by their own description a “bastard” Portuguese, including French words and colloquialisms, this
variety is intolerable to purists, but understood by the 900,000 strong immigrant
population. In Albatroz, a literary review trilingual in Portuguese, French and
immigrais, the proponents of this variety write (Muñoz, 1999:71):
Example 3
Through no will of our own, we’re foreigners. But we have a tool: the Portuguese
language contaminated by positive pollution. Be that as it may, we write in both languages, copulating frenetically, with the outcome that literary and pictorial objects are
pleasantly produced. It’ll be alright. We respect none of the spelling or vocabulary rules of
the immigrese language and although we may perturb, we demand that making mistakes
be recognized as a technique for exploring the ambiguity of a text or as a gimmick of
polysemic amplification. Don’t you agree? And while we’re at it, let’s do away with the
cedilla! Irreverence, dear reader, irreverence.4
Hewitt (1986) and Sebba (1993) have described types of CS which occur
with symbolic and discourse-related functions in London, between Creole and
London English speaking young people. Although used sparingly, creole is said
by Sebba to have a we-code function and CS is described as an “insider
activity”. Creole is almost certainly preserved by being used in this way, as
the majority of young people who were the subject of his study would not be
able to use it as a means of expression on its own. Recent novels such as White
Teeth by Zadie Smith, which portray the lives of young Afro-Caribbeans in
London, show this type of CS as an intrinsic – if somewhat inconsistent – aspect
of their speech.
Finally, there are an increasing number of popular singers and bands whose
lyrics are code-switched, e.g. Ricki Martin, Raggasonic and various Punjabi
bands in Britain (Asian Dub Foundation, Bali Sagu, Apache Indian), attesting
the vitality which code-switched varieties can have in their own right. Example
4 is taken from the lyrics of Raggasonic Crew Feat,5 sung by Raggasonic, a
Rasta band who sing in English, French and a mixture of both. Along with
English–French CS, their lyrics contain morphologically adapted words, borrowed from one language to the other (e.g. toaster); elements of creole; and
verlan (shown in example 4 in capitals), a type of French slang which involves
inverting the first and second syllables of multisyllable words and pronouncing
single syllable words backwards (from envers ‘back to front’).
Example 4
1 Les femmes sont nice
2 et la dancehall roule
3 La basse percute tous les Rastas
4 dans la foule
5 Le sélécteur joue du rub-a-dub
6 ou de la soul
7 Les bad-boys sont là
8 Les bad-boys sont là
9 Eclaire le chemin
10 Quand Raggasonic SSEPA [=passe]
11 La dancehall est full,
12 Mory, Big Red sont vraiment là
13 Je viens d’Ivry, le Sud
The women are nice
and the dancefloor is rolling
The bass hits all the Rastas
in the crowd
The selector plays some rub-a-dub
or some soul
The bad-boys are here
The bad-boys are here
Light the way
When Raggasonic goes by
The dancefloor is full
Mory, Big Red are really here
I come from Ivry, the South
My translation.
The transcription is taken from the lyrics attached to the CD. The translation is my own.
Code-switching and language contact
14 est le best dans le ragga
is the best at reggae
15 Si t’es pas d’accord bwoy6
If you don’t agree bwoy
16 viens donc nous CHECLA
come on and clash us
[= clasher, i.e. ‘to clash’, adapted to French spelling and morphology]
17 J’ai pas peur de la compétition
I’m not afraid of the competition
18 j’ai pas peur de toi
I’m not afraid of you
19 Tu peux toaster comme Buju
You can toast like Buju
20 ou tester comme Shabba
or test like Shabba
21 Toaster comme Beanie Man
Toast like Beanie Man
22 ou bien comme Bounty Killa
or else like Bounty Killa
23 Je n’ai peur d’aucuns deejays
I’m not afraid of no deejays
24 des Antilles à RIPA [=Paris]
from the Antilles to Paris
25 People massive écoute ça
massive people listen to that
Here, prescriptive “rules” such as speaking only one language at a time are
deliberately and playfully broken – a function of CS which has been described
as ludic (Caubet, 2001; McCormick, 2002). The mixing is not a precursor of
any of the varieties involved disappearing, at least not among the writers and
consumers of such lyrics as these, who are fluent multilinguals – rather it is the
purists who should feel threatened! The various functions of CS in rap music are
explored in Sarkar and Winer (2006) and in North African-French rai music by
Bentahila and Davies (2002). Both claim that one of its functions is to reconcile
the conflicting trends of localization and globalization.
CS can therefore occur both in situations of decline and as a mechanism of
vitality. Aikhenvald (2002:Chapter 8) shows how an Amazonian community,
the Tariana, have a strong inhibition against any kind of mixing of their
language with that of other local tribes, such as the Tucano, who are gaining
dominance over them, and the Makú, whom they despise, comparing them to
dogs and accusing their language of containing “inhuman sounds”. When such
mixing occurs, it evokes ridicule and pity. However, they are increasingly
allowing CS with Portuguese and even with English, which symbolizes “everything a capitalist paradise could offer” (2002:211). Gibbons (1987) describes a
variety called ‘MIX’ spoken by students at Hong Kong University, principally
among themselves. Unlike a creole (see below), MIX has emerged within a
single community and the processes which have occurred have taken place
entirely within this relatively homogeneous group.
In reviewing research which shows the variety of circumstances where
CS may arise, Heller suggests as a priority for CS research to look at “the extent
to which different types of CS are related to different types of boundary
maintenance/change processes” and “the generalizability of findings concerning
In lines 6 and 11, phonological CS serves an artistic purpose: by pronouncing soul [su:l] and full
[fu:l] they rhyme with the French words roule (line 2) and foule (line 4). In line 15, bwoy is creole.
the social conditions under which CS is or is not found” (1988a:268–269). More
comparative research is needed (cf. Appendix) for us to realize this aim.
Code-switching in language interaction
The search for regular correspondences between language interaction and its
accompanying social factors has a relatively short history. Traditionally it was
thought that change followed universal, language-internal principles such as
simplification, and therefore took place in the absence of any contact with other
varieties (J. Milroy, 1998).
Equally, CS has not always been considered an aspect of language change.
Data from the Puerto-Rican community in New York (Poplack, 1980; 1983)
was used to argue that. CS was simply the alternation of two varieties, English
and Spanish, which preserved their monolingual characteristics. CS was opposed
to borrowing, which was seen as a form of convergence (see Chapter 5). On the
other hand, Mougeon and Beniak (1991:9) suggested that Poplack downplayed
the role of CS as an explanation of language change, as a reaction to exaggerated
claims that immigrant varieties are hybrid and their speakers inferior. They
themselves set out to “rehabilitate” the role of interference in language change,
refusing to see the term as pejorative. CS is viewed as the most likely source
of borrowings (see Chapter 1). The exact role of contact in language change
is, however, still very much a matter of discussion (Clyne, 2003; Harris and
Campbell, 1995).
Code-switching, interference and borrowing
Romaine (1989) and Myers-Scotton (1992) reviewed the extensive discussions
in the literature concerning the relationship between borrowing and CS, which
was mentioned as a significant issue in Chapter 1. One reason why this question
has been raised so often is that single-word code-switches/loans are, in many
situations, though not always, the commonest kind of CS. Any aspect of a
language, however, including its structures, can be borrowed (Boeschoten, 1998).
Within single words, common nouns are the most frequently borrowed
items (Poplack, Sankoff and Miller, 1988:62). One explanation of this is provided by Bynon (1977:231), who says this just reflects the size of the grammatical categories concerned. Another is suggested by Aitchison (2000:62), who
points out that nouns are freer of syntactic restrictions than other word-classes.
Both these explanations involve falling back on “language-internal” factors
rather than sociolinguistic ones. Another reason for the prevalence of singleword switches/loans is that these are accessible to bilinguals with any degree of
competence, even minimal, in the language from which the borrowing is taken.
It was argued in Chapter 1 that there is no reliable way of distinguishing
Code-switching and language contact
synchronically between loans and code-switches; loans must start life as codeswitches and then generalize themselves among speakers of the borrowing
language (Haust, 1995). When this happens at different historical stages of
contact between the same two languages, the loans may go through quite different processes of integration and end up looking quite different in the receiving
language. Heath (1989), for example, has shown how some French verbs,
borrowed in the early French colonial period, were adopted in Moroccan Arabic
without inflectional verb frames. Others, borrowed more recently, have been
instantly provided with Moroccan Arabic inflectional frames. Their phonemes
have also been imported wholesale, and show signs of stabilizing as such
Sankoff (2001) descibes CS as the “royal road” to borrowing although,
somewhat paradoxically, she still supports Poplack’s idea that there is an
essential distinction between the two. The distinction, it is claimed, is demonstrated by the fact that loan words and “ambiguous lone items” show different
statistical patterning in the corpus of Poplack and Meechan (1995). This difference, however, could be a matter of how the two concepts have been defined for
research purposes – Sankoff claims that it is preferable to study borrowing
processes independently from the “muddy waters” of CS. This preference for
studying categorical – and preferably countable – phenomena leads us back to
the argument about essentialism (Chapter 1). Grammatical category Nouns may be the most frequently borrowed –
and switched – word-class owing to their grammatically self-contained character,
but all grammatical categories are potentially transferable. In some data-sets,
other types of CS are more frequent. In a comparative study of Punjabi–English
bilinguals and Greek Cypriot–English bilinguals, intra-sentential switching was
almost three times as frequent as single-word switching among the Punjabis,
whereas single-word switching was the commonest among the Greek Cypriots
(Cheshire and Gardner-Chloros, 1998). The Punjabi speakers switched massively
more than the Greek Cypriots overall, so a possible hypothesis is that the more
CS there is overall, the smaller proportion single-word switching represents.
Clyne (2003) has rightly pointed out that it would be worth investigating the
role of typological factors also in this situation – although at first glance, it should
be easier to produce complex switches between Greek and English than between
Punjabi and English. Morphophonemic integration with the surrounding language Codeswitches as well as loans can be morphophonemically integrated with the borrowing language. Borrowed/code-switched verbs frequently take the morphology
of the borrowing variety, e.g. Alsatian déménagiert ‘moved house’ from French
déménager (Gardner-Chloros, 1991); Maori changedngia ‘to change’ (Eliasson,
1991); Spanish coughas ‘you cough’ (Zimman, 1993); German gedropped
‘dropped’ (Eppler, 1991); etc. Native synonym displacement There are examples of both loans and
code-switches filling lexical gaps in the borrowing language, and of their
adding themselves as a further option to the native equivalent. The fact that
CS is not always the result of an inability to find the right word or expression,
is demonstrated by the fact that a common conversational function of CS is
repeating (more or less exactly) what one has just said in the other language.
Example 5
r e n u : and she was sleeping all over the place, so I had to stay awake
so I had to stay awake
she was falling around everywhere, so I had to stay awake
(Gardner-Chloros, Cheshire and Charles, 2000:1319)
The speaker is clearly fluent enough to have the relevant expression available
to them in both varieties. The repetition is functional in terms of its effect within
the discourse, breaking up the monotony of repeating the whole expression
in English (Gardner-Chloros, Cheshire and Charles, 2000). Similarly, bilingual
parents issuing brief orders to their children (“Come!”, “Put it on!”) often both
soften and reinforce the instruction by repeating it in the other language. By
contrast, in mot juste switching (Poplack, 1980), speakers switch precisely
because the other language contains the most accurate term.
Example 6
No me precipitaré en el famoso name-dropping
I will not throw myself headlong into the famous practice of name-dropping
(McClure, 1998:134)
Switching of this type may lead to the word becoming a fully fledged loan,
although this never depends on linguistic factors alone, but also on sociocultural ones.
This was, for example, the conclusion reached by Treffers-Daller (1994). First
of all she found, in her study of French–Dutch contact in Brussels, that the two
varieties shared numerous phonemes: many words such as unique and sympathique could belong to either variety. Second, both French and Dutch have
limited morphological marking, so that the criterion of morphological integration could often not be used. Third, both borrowings and code-switches varied as
to whether they were syntactically integrated or unintegrated. Treffers-Daller
therefore concluded that a unified theory of borrowing and CS was needed.
Those who disagree would point to the fact that in a given corpus, loans and
code-switches may be clearly distinguishable. For example, in Jones (2005),
speakers of Jersey French (“Jèrriais”) tended to flag code-switched forms (with
pauses, self-corrections, etc.), but not borrowings; furthermore those informants
Code-switching and language contact
with the most positive attitude towards Jèrriais avoided CS entirely (see also
Poplack, 1988, discussed in Chapter 3). It is clear that even where a continuum
between linguistic varieties exists, speakers are able to use different points
along it in a contrastive manner (phonological examples spring to mind). This
is different from the question of whether those phenomena are intrinsically,
or invariably, distinct.
Code-switching and pidginization/creolization
CS is found alongside pidginization and creolization/decreolization in many
parts of the world, and contributes, as they do, to the convergence and divergence of different varieties. Gumperz (1964) and Gumperz and Wilson (1971)
identified the close relationship between CS and creolization in the early days of
CS research; more recently the trend has been to emphasize the differences.
Differences there clearly are, but this should not distract us from the fact that
these processes often co-occur, derive from similar social factors and may
sometimes lead to similar outcomes.
Crowley (1990), for example, argues that in Melanesian Pidgin, access to the
substratum (i.e. indigenous languages) persists at all levels: stabilization, destabilization and creolization should all, he claims, be regarded as contributing
simultaneously to the gradual evolution of the language (pp. 384–386). As to CS,
its presence or absence seems to be largely a function of the prestige attached to
the pidgin. Comparing usage on the radio in Vanuatu and the Solomon Islands,
Crowley finds much more CS in the Solomon Islands, where the pidgin’s prestige
is low, than in Vanuatu, to such an extent that in the case of the Solomon Islands
announcers, “It is sometimes difficult to know which language they would claim
to be speaking.”
In Papua New Guinea, Romaine (1992) shows how CS occurs within the
context of a post-creole continuum which had emerged in the preceding twenty
years: “In town, Standard English, English spoken as a second language with
varying degrees of fluency, highly anglicized Tok Pisin, more rural Tok Pisin
of migrants, and the creolized Tok Pisin of the urban-born coexist and loosely
reflect the emerging social stratification” (p. 323). Whereas some have been
adamant that this situation should be seen in terms of ongoing CS between Tok
Pisin and English (Siegel, 1994), Romaine believes that there is no principled
way for determining to which language many utterances belong to (1992:322).
Le Page and Tabouret-Keller (1985) give examples of creole speech from
sources in the Caribbean, Belize and London. At its most fluid, CS involves
shifting at particular linguistic levels rather than a wholesale transition from
one variety to another – indeed discrete codes are hard to come by in linguistic
contexts as unfocused as those they describe. Sometimes, as in the case of London
Jamaican, the switching between the codes is more symbolic than real:
London Jamaican is more a set of norms to be aimed at than an internally coherent
and consistent system. Speakers behave as if there were a language called “Jamaican”,
but often all they do (perhaps all they know how to do) is to make gestures in the
direction of certain tokens associated with Jamaican Creole which have a stereotypical
value. (1985:180).
Finally, one linguistic phenomenon which shows what a fine line there is
between creolization and CS, is bilingual compound verbs (Muysken, 2000;
Myers-Scotton, 2002a; Edwards and Gardner-Chloros, 2007). Romaine (1986)
showed how in English–Punjabi CS, Punjabi verbal “operators” meaning ‘do,
make’ are commonly combined, in the speech of bilinguals, with a major
category (noun, verb, adjective) taken from English, to make new verbal
compounds which function as a single syntactic/semantic unit, e.g. ple kerna,
where ple is from English play and kerna means ‘to do/make’ in Punjabi. The
new compound, which means ‘to play’, is synonymous with an existing Punjabi
verb. Parallel creations have been attested in other, typologically diverse,
language combinations, including some, such as Greek Cypriot–English,
where, unlike Punjabi, neither language provides a native model for the compound verb formation, e.g. kamno use, kamno respect, kamno developed,
kamno spelling, where the Cypriot form kamno ‘make/do’ is combined with
various English words, to make new verbs. These compounds appear superfluous in that the same meaning could in each case be conveyed by a single
equivalent in Greek (Gardner-Chloros, 1992). Although generally considered
under the heading of CS, these formations show features of creolization, as they
involve grammatical convergence and an analytic approach to vocabulary. Heath
(1989) gives examples from French borrowings into Arabic and Arabic borrowings into Turkish. These are interesting because, as Heath points out, they occur in
intensive contact zones with alternative, productive adaptation routines. Along
with the avoidance of inflections reminiscent of pidgins, some of these importations also show unstable word-class, such as himri in Moroccan Arabic (from
English hungry), which fluctuates between noun and adjective (1989:202).
This also may be viewed as a product of the fuzziness of the borders, in
linguistic terms, between different types of contact languages. But whereas
prototypical pidgins and creoles arise as lingua francas in situations involving contact between several languages, other types of mixed languages arise
where there is widespread bilingualism and where the new variety serves
mainly as a marker of ethnic group identity (Thomason, 2001:197).
Code-switching and mixed languages
“Mixed languages” are generally thought of as a limited and marginal group of
languages, spoken in very specific contexts. Their social genesis is crucial to
understanding their grammatical make-up. Unlike pidgins, they do not arise
Code-switching and language contact
in circumstances where there is no common language, but rather as the outcome
of particular cases of bilingualism.
They are of interest here because they appear to represent a kind of fossilization of CS (Bakker and Mous, 1994; Bakker, 1997). They characteristically
have the grammar of one language and the lexis of another, but in practice this
split is “rarely if ever consistent” (Matras, 2000a). Hamers and Blanc (2000) and
Thomason (2001) have discussed the differences between CS, mixed languages
and pidgins and creoles. Some mixed languages show much more specific and
regular patterns than is generally the case with CS. Michif, for example, spoken
by the Métis in western Canada and North Dakota and described in Bakker
(1997), has a split grammar. The structure of the nominal phrase is essentially
French, and that of the verb phrase essentially Cree. A more doubtful candidate
for membership of this group is Chiac. A variety of Acadian French–English
spoken principally by young people in Moncton, Canada, Chiac is age-graded
and highly variable. It can perhaps best be described as a slightly conventionalised type of CS (Perrot, 2001).
As Matras says, most languages are mixed to some extent, and although
mixed languages represent a particular type of focusing, the processes underlying them are not fundamentally different from those operating in other forms
of contact. The motivation underlying mixed languages, as Bakker points out
in relation to Michif, is probably similar to the motivation for CS: the linguistic
expression of a dual identity. English could be seen as a mixed language,
consisting of an Old English grammatical base with partial relexification from
Norman French. “At first those who spoke French were those of Norman origin,
but soon through intermarriage and association with the ruling class numerous
people of English extraction must have found it to their advantage to learn the
new language, and before long the distinction between those who spoke French
and those who spoke English was not racial but largely social” (Baugh,
1951:135). In line with Bakker’s (1997) account of mixed language formation,
there were numerous marriages between Norman men and English women
(Baugh, 1951:141), which is consistent with Norman being the lexifier and
Old English the principal provider of grammatical structure. Most striking of all
is the evidence that the introduction of French words into English is closely
correlated with the progressive adoption of English by the upper classes.
According to Jespersen (1928:94), slightly over 10,000 French words entered
the English language between 1250 and 1400.7 This corresponds exactly to the
period when the upper classes were adopting English (these words represent
40 percent of all the French words in English, of which about 75 percent are still
in current use). From the fifteenth century onwards, when English had become
The calculation is based on these words’ first recorded usage.
the language of the majority within the ruling classes, there was a sharp drop in
borrowings from French (Jespersen, 1928:94; Baugh, 1951:214).
Structural v. social influences
Since the work of Thomason (1986) and Thomason and Kaufman (1988), it
has become accepted that many aspects of language change are due to contact –
and conversely, that contact often gives rise to change over a longer or a shorter
period of time. Chambers and Trudgill (1999) showed that many changes arise
from contact between different dialects of the same language, so what appear
to be internal developments are often due to processes such as inter-dialect
accommodation, imperfect learning and koineization (i.e. the creation of a
common variety from existing ones). Trudgill set out “to link a typology of
languages and language change to a typology of societies” (1997:3). To do this,
he sought out isolated dialects and languages so as to discover which changes
arise, and why, in the absence of any contact. Thomason and Kaufman also
aimed both to systematize the linguistic factors and to make generalizations
about the social influences which affect them (1988:36). The quotation above
from Heller set out similar aims specifically relation to CS.
The novelty of Thomason and Kaufman’s approach lies in their conviction
that social factors determine fundamental aspects of language change. “It is the
sociolinguistic history of the speakers, and not the structure of their language,
that is the primary determinant of the linguistic outcome of language contact”
(1988:35). As in other taxonomies of language interaction, they divide the
outcomes of contact into categories: the first is borrowing, which can affect
both the influenced and the influencing variety, and which, they claim, results
first in the adoption of lexical elements. Extensive structural borrowing, on the
other hand, is a one-way process which requires extensive bilingualism among
the borrowing-language speakers. Another outcome is interference through
shift, which does not begin with vocabulary, but with sounds, syntax and
morphology, and occurs in the absence of full bilingualism.
Thomason and Kaufman do not specifically discuss CS, but their claims
are highly relevant to it owing to their emphasis on the asymmetric quality
of contact, which links it to the social context within which it occurs. The idea
that it is sociolinguistic rather than structural – and potentially universal –
factors which determine the outcome of language contact is relevant to the
proposed universal grammatical constraints on CS (Chapter 5). In the past,
linguists similarly tried to establish universal rules governing the outcome of
“interference” generally. For example, Thomason and Kaufman quote various
authors who claimed that contact-induced change results in simplification of
the structures of the contact-receiving language. They then show, by means of
counter-examples, that the reverse can also be the case. For instance, contact
Code-switching and language contact
between the Cushitic language Ma’a and Bantu led to the wholesale adoption by
Ma’a of highly marked Bantu inflectional structures. Second, Asia Minor Greek
adopted morphological, phonological and syntactic features of Turkish. Their
explanation for the latter is that it was the Greeks who were under cultural
pressure, and who therefore became bilingual. The Turks, for their part, merely
borrowed from Greek at a lexical level. As we saw, the minority language is
likely to be significantly influenced by the majority one rather than the other
way round. Just as there is usually an imbalance in power/status between social
groups in contact, so the resulting CS reflects this imbalance and is itself nonsymmetrical. Johanson (2002) avoids the terms CS, borrowing and interference
and instead discusses “code copying”, which is neither a clear switch from a
basic code to a foreign code nor a fusion of the two. He proposes that some
linguistic features are more “attractive” than others and thus more likely to be
borrowed. But the concept of “attractiveness” is not clearly operationalized and
Johanson himself agrees with Thomason and Kaufman that “social factors
ultimately determine the extent to which attractiveness leads to influence”
(2002:3; see also the discussion in Backus, 2004).
Structural factors still have a role to play. Heine and Kuteva (2005) argue
that “in situations of intense language contact, speakers tend to develop some
mechanism for equating ‘similar’ concepts and categories across languages”.
However the speakers’ perception of equivalence may not be the same as the
linguist’s and the course of contact is not always smooth or obvious. For
example, Heine and Kuteva claim that it does not always result in simplification, structural parallelism is not always favoured, and, as Tsitsipis (1998)
points out, changes may not always be completed, and may be continuous or
An example of the interplay of structural and social factors is provided by
Treffers-Daller (1994; 1999), who compared CS between French and Brussels
Dutch in Brussels with that between French and the Alsatian dialect in
Strasbourg (using data in Gardner-Chloros 1991), reaching slightly different
conclusions in the earlier and the later study. The impetus for the comparison
was the fact that both in Brussels and in Strasbourg, the local varieties of a
Germanic language are in contact with French, from which they have borrowed
extensively. However, whereas a mixed identity, relating both to French and
Alsatian, is the norm in Strasbourg, in Brussels, where the two linguistic groups
are more polarized, all except the oldest members of the population identify
either with French or Dutch, but not with both. Consequently, as TreffersDaller points out, Brusselers “no longer consider the mixed code to be an
appropriate expression of their identity” and intra-sentential CS is accordingly
In the later paper (Treffers-Daller, 1999), the differences between CS in the
two situations are downplayed, and it is argued that the typological similarities
between the two contact situations are determinant and lead to similar outcomes despite sociolinguistic differences. However, in both these instances
of French–Germanic contact, French is the more prestigious language, and
this too might help explain certain similar outcomes in terms of structural
transfer. Further systematic comparisons will be necessary in order to
establish with more certainty the relative role of structural and sociolinguistic
The examples below are first from Brussels and second from Strasbourg.
In both, the Germanic sentence structure with the main verb at the end provides
the template for the mixed utterance:
Example 7
(a) Surtout ze hebben een brief gemaakt.
Above all they have a letter made
Above all, they have made a letter.
(Treffers-Daller, 1994:192)
(b) Wäje
dem han se faillite
Because of that have they bankruptcy made
Because of that, they went bankrupt.
(Gardner-Chloros, 1991:131)
In both cases, the Germanic variety has imposed its characteristic word order
on the code-switched sentence, with the direct object placement before the main
verb, which comes at the end. Treffers-Daller does not claim this as a universal
feature of French–Germanic contact, and in fact contrasts this situation with
that described in Clyne (1987). The latter found evidence of this structure
giving way to English word order in Australian Dutch as a result of the impact
of English, which exerts a powerful sociolinguistic influence on the minority
Nonetheless, Treffers-Daller’s findings require some explanation. One possibility is that, in spite of some differences, the two situations are still sufficiently
similar at a sociolinguistic, as well as a linguistic, level for the outcome of
language contact to be similar, whereas in Australia the situation is much more
different than it is between these two. Another possibility is that, although the
two linguistic situations differ in many respects, they are nevertheless broadly
similar in terms of the overall balance of the linguistic vitality of the Germanic
and the Romance variety. This could explain a similar linguistic outcome despite
some significant differences of detailed sociolinguistic distribution. Bearing in
mind that CS is an unstandardized form of speech, a third possibility is that
idiolectal factors intervene. There is in fact a difference between example 7a,
where the verb–subject order is not inverted as Dutch would require (ze hebben),
whereas in the Alsatian case it is (han se). Internal variation and/or inconsistency undoubtedly arises in both communities. In Alsace, structures based on
an Alsatian “template” can be found in French as in example 7, but the opposite
can also be found, e.g.:
Code-switching and language contact
Example 8
Es saat mir nix
It tells me nothing
[It doesn’t appeal to me, calqued from the French expression Ça ne me dit rien].
(Gardner-Chloros, 1991:177).
In general terms, the Alsatian situation supports Thomason and Kaufman’s
proposals: Alsatian borrows extensively from French at a lexical level; some
Alsatian words are also borrowed into French, but less than four times as often
(Gardner-Chloros, 1991:164). This assymetry is in accordance with their
predictions (Thomason and Kaufman, 1988) as to the likely outcome of shift
from a low-vitality to a high-vitality language. “Contact-induced shift”, the
other outcome predicted in the model, whereby imperfect learning of the target
language results in alterations to that language (here French) of a mainly structural
nature, can also be found, e.g.:
Example 9
Toujours il était bon
Always it was good
It was always good.
(Gardner-Chloros, 1991:177).
The Alsatian word order, with the adverb placed at the beginning of the
sentence, is used (instead of French: il était toujours bon) although there is no
inversion of the verb and subject as required in Alsatian (see example 7b above).
The outcome of “incomplete” Alsatian influence on French is therefore a hybrid
In some cases, sociolinguistic factors can be seen to override the structural
closeness/distance between the languages. We saw above that CS patterns in two
broadly comparable immigrant communities, the Greek Cypriot community
in London and the Punjabi community in Birmingham, diverged widely: “On
every count which we were able to compare, the Punjabis switched massively
more than the Cypriots and more intra-sententially than through any other
form of switch … Whereas there were under five intra-sentential switches per
ten utterances for the Cypriots, there were over sixty for the Punjabis” (Cheshire
and Gardner-Chloros, 1997:270). This is the reverse of what one would expect
on the basis of structural closeness/distance between the two varieties, Greek
and English being typologically closer than Punjabi and English.
CS is one of the possible outcomes of contact between two (or more) varieties,
often co-existing and overlapping with other outcomes. Owing to the huge range
of linguistic guises which it adopts, it has sometimes been ignored altogether in
studies of language contact, and sometimes defined as being more neat and tidy
Box 3 Medieval Code-Switching
Many code-switched texts, both official and non-official, survive from
medieval times (see Trotter, 2002). These cover a variety of genres, including sermons, medical texts and business documents. In Britain, English was
code-switched with Latin but also with Norman French. Schendl (2002:81)
quotes the following letter, from the Dean of Windsor, Richard Kingston, to
King Henry IV (1403). He points out that the use of inter- and intrasentential
CS in a letter to the king indicates the social acceptability of code-switching
at the time.
Letter from R. Kingston to King Henry IV (1403)
Please a vostre tresgraciouse Seignourie entendre que a-jourduy apres
noone…qu’ils furent venuz deinz nostre countie pluis de .cccc. des les
rebelz de Owyne, Glyn, Talgard, et pluseours autres rebelz des voz marches
de Galys … Warfore, for goddesake, thinketh on your beste frende, god,
and thanke hym as he hath deserved to yowe! And leueth nought that ye ne
come for no man that may counsaille yowe the contrarie … Tresexcellent,
trespuissant, et tresredouté Seignour, autrement say a present nieez. Jeo
prie a la benoit trinité que vous ottroic bone vie ovc tresentier sauntee a
treslonge durré, and sende yowe sone to ows in help and prosperitee; for in
god fey, I hope to almighty god that, yef ye come youre owne persone, ye
schulle have the victorie of alle youre enemyes … Escript a Hereford, en
tresgraunte haste, a trios de la clocke apres noone. le tierce jour de
May your most gracious Lordship be pleased to hear that today, in the afternoon ...
more than 400 of Owen, Glyn and Talyard’s rebels, and several other rebels from your
Welsh borders, entered our county. Wherefore, for God’s sake, set your mind on God
as your best friend, and thank him for the favours he has bestowed upon you. And do
not for any reason fail to come, whatever advice to the contrary you may receive from
anyone … Most excellent, most powerful and most redoubtable Lord, let me be denied /
refused in some other way! I pray the blessed Trinity that you be granted good life
with perfect health for a long time to come, and may [the Trinity] send you to us soon in
help and prosperity; for I faithfully pray to almighty God that, if you yourself come in
person, you will be victorious over all your enemies … Written [by R. Kingston to King
Henry IV] at Hereford in the utmost haste at three o’clock in the afternoon on the third day
of September [1403].
NB: In the original edition, the words please, noone and clocke were
italicised as if they were English, but they are in fact French words of
French origin (just like deserued, counsaille, prosperitee, etc.), so they are
not italicized here in the original or in the translation.
Code-switching and language contact
than it actually is. There are examples of relatively stable bilingual situations
where two varieties appear to alternate without affecting one another’s essential
character, but in other cases CS is rule-breaking behaviour, which should
be seen not in relation to static norms but in terms of language change and
convergence. A fuller exploration of the mechanisms by which the latter occurs
is called for (see the papers in Bilingualism, Language and Cognition, 7(2),
August 1974). Toribio, for example, explores the different factors involved in
convergence, and argues that the “semantics–pragmatics interface” is crucial
(2004:167). She agrees with Otheguy (1995) that bilinguals select “the most
parsimonious grammar that serves both languages”. Bullock and Toribio (2004)
distinguish it clearly from interference and transfer; rather than implying the
imposition of a structural property from one language on another, they see it as
an “enhancement of inherent structural similarities found between two linguistic
systems” (p. 91). It is to be hoped that the relation of CS to these processes will
continue to be investigated.
It increasingly appears that sociolinguistic factors are the key to understanding why CS takes the form it does in each individual case. Such factors
affect different sub-groups in different ways, so there are often different types
of CS within the same community. At a social level, CS may be seen as the
product of a power struggle between two varieties (Pujolar, 2001) (see Chapter 3).
At an individual level, it reflects varying bilingual competences and serves as
a discourse-structuring device. Its functionality is discussed in Chapter 4, which
reinforces the view that CS is not a passive victim of linguistic forces.
Social factors in code-switching
In Chapter 2, we saw that CS is often manifest within the process of language
change, which can lead to the creation of new varieties such as pidgins or mixed
languages. In other cases, it may be a temporary phenomenon, leading only to
some limited borrowing. The outcome of language contact situations is determined
by social and economic variables: the relative prestige of one variety as opposed to
another, or its association with a more powerful or up-and-coming group. In this
chapter, we will look at social factors in CS – factors which, as we saw in Chapter 2,
are as important as, if not more important than, the linguistic characteristics of the
varieties in determining the linguistic outcome. Whether used in a deliberate way,
as above, or not, CS provides a variety of clues as to the social identity of the
speaker – the groups which, to paraphrase Le Page, she or he wishes to resemble. In
Chapter 4, we will see that bilingual speakers often use CS as “conversational
scaffolding” while at the same time using it to convey aspects of their identity. The
motivation to code-switch relies on factors independent of the varieties as such,
including the speakers’ relative competence and that of their interlocutors, the
identities they can express through each language, the acceptability of CS in their
network and in particular contexts, and a variety of further factors.
Types of factor
As Thomason and Kaufmann pointed out (see Chapter 2), there are a range of
factors which determine whether or not CS occurs at all in a given language
contact situation. From a sociolinguistic point of view, three types of factor
contribute to the form taken by CS in a particular instance:
(1) factors independent of particular speakers and particular circumstances in
which the varieties are used, which affect all the speakers of the relevant
varieties in a particular community, e.g. economic “market” forces such as
those described by Bourdieu (1997), prestige and covert prestige (Labov,
1972; Trudgill, 1974), power relations, and the associations of each variety
with a particular context or way of life (Gal, 1979).
Social factors in code-switching
(2) factors attaching to the speakers, both as individuals and as members of a
variety of sub-groups: their competence in each variety, their social networks and relationships, their attitudes and ideologies, their self-perception
and perception of others (Milroy and Gordon, 2003).
(3) factors within the conversations where CS takes place: CS is a major
conversational resource for speakers, providing further tools to structure
their discourse beyond those available to monolinguals (see Chapter 4).
There are many overlaps and inter-relations between the three sets of factors, and
some understanding of all three is necessary in order to apprehend why particular
CS patterns arise. The classification provides a semblance of order within the huge
range of factors which attach neither to the varieties themselves as linguistic
entities, nor to cognitive/psycholinguistic factors which affect the individual. For
example, the individual’s competence in the relevant varieties is a product of their
(reasonably permanent) psycholinguistic make-up; at the same time, it has sociolinguistic implications, as it is closely connected with factors such as age, network
and identity. Thus, whether or not a second- or third-generation member of the
Chinese community on Tyneside can converse fluently in Chinese determines
the extent to which they can take part in conversations with the oldest members of
the community, who may be to all intents and purposes monolingual Chinese
speakers. At the same time, their social networking with people their own age is
also partly determined by their linguistic abilities, and their association with
English or Chinese speakers is likely to reinforce their preferences and abilities
in those languages (Milroy and Li Wei, 1995; Li Wei, 1998a).
Code-switching in two communities: Strasbourg
and London Cypriots
Two examples based on field-work in Alsace (Gardner-Chloros, 1991; 1997)
and among Greek Cypriots in London (Gardner-Chloros, 1992; 1995; GardnerChloros, McEntee-Atalianis and Finnis, 2005) are presented below, in order to
illustrate how these various factors interact in real-life situations. The two cases
contrast in that one is a native plurilingual and the other an immigrant setting.
The main similarities are that both settings are European, and the varieties
involved all Indo-European.
In order to assess the contribution of the linguistic and sociolinguistic factors
properly, one would need to systematically compare a variety of situations
where CS occurs. For example, one would need to contrast situations where
there is switching between closely related varieties with other situations where
the varieties involved are typologically far apart from one another. Similarly,
one could try to assess the importance of sociolinguistic factors by contrasting
cases where the social situation is similar with others where it is quite different.
Some attempts to evaluate the relative weight of these aspects have been made
(see below). The purpose of the LIPPS/LIDES enterprise is partly to permit such
comparisons, by setting up a database of CS texts (see Appendix).
Strasbourg Background Strasbourg is the capital of Alsace, the easternmost
region of France, and a historic town of some quarter of a million inhabitants.
Originally part of the Germanic Holy Roman Empire, this disputed region
changed hands between France and Germany five times between 1648 and
1945. Since the last return to France after the Liberation, a clear French identity
has prevailed. French is the language of the state, of education and the media
and, increasingly, the mother tongue of the younger generation.
Surveys have revealed that the Alemannic dialect known as Alsatian,
which is on a dialectal continuum with dialects spoken in the adjoining
areas of Germany, is still widely spoken by various sectors of the population
(Gardner-Chloros, 1991; Schuffenecker, 1981; Tabouret-Keller and Luckel,
1981a and 1981b). The existence of this dialect has always been an important
element of the area’s identity, particularly during the times when it formed
part of France rather than Germany. It is used more in rural than in urban
contexts, and varies with the age and gender of the speaker, being spoken
most by the sociolinguistic groups where one would expect a traditional
dialect to be used, i.e among older, rural males. In Strasbourg, it also functions
as an in-group marker of Alsatian identity, as opposed to the overarching
French identity (French people from other parts are known as “les Français de
l’Intérieur” in Alsace). The dialect also distinguishes locals from the international community associated with European institutions based in Strasbourg,
whose staff may speak French, but very rarely Alsatian. Alsatian was rigorously avoided after the Second World War for its Germanic associations,
but has recently enjoyed a revival among younger generations, including
young parents when speaking to their offspring. It is used in adolescent
groups as well as by local comedians, playwrights and poets, for humorous,
emotional and vernacular purposes of all kinds. There is even a version
of Microsoft Office currently being translated into Alsatian by a team of
In spite of this, the surveys document a slow but inexorable falling off in its
use as one goes down the age groups. Bister-Broosen’s more recent research
(2002) identifies positive attitudes towards the dialect among the young generation, but points out that such positive attitudes are common in the final stages
of language loss and that they should therefore in fact be interpreted as a
warning of the dialect’s impending disappearance. In Strasbourg, Alsatian is
more often than not mixed with French, i.e. it is heard more in the context of
code-switching than in its “pure” form.
Social factors in code-switching
45 Types of code-switching The following example is typical of CS as
heard in Strasbourg. It represents the full text of a telephone conversation between
a woman in her thirties working in an insurance office, Mrs F, with a technician
who had already tried (unsuccessfully) to repair the air-conditioning system.
Example 1
1 mrs f:
Do isch d’Madame F. De bureau de Saint-Paul. Bonjour. Wissen’r
warum ich schunn widder anruef? Unseri mache schunn widder uf siewene
zwanzig Grad … Jo, c’est impossible, hein. Madame Jund, quand elle
est arrivée ce matin, il y en avait 29 … Do het se Tüer e bissel ufgemacht.
Mir kann awwer ken … ich weiss nitt. Ce gars était là l’autre jour un
dann het er ebbs gemacht, het g’saat … et puis tout d’un coup ça a fait
clic et effectivement ça a remarché; vendredi il faisait bon, angenehm
et tout. Et aujourd’hui ça va être pas tenable parce que’s gibt ken Luft,
hein … Siewene zwanzig Grad, isch e bissel viel … Ja, j’compte sur
vous, hein. Merci Monsier Raab, au revoir.
It’s Mrs F here. The Saint-Paul office. Hello. Do you know why I’m
ringing up yet again? Our (thermometers) have gone up again to twenty
seven degrees … Yes, it’s impossible, eh. Mrs Jund, when she arrived this
morning, it was 29 … So she opened the door a bit. But we can’t … I don’t
know. This guy was here the other day and then he did something, he
said … and then all of a sudden it went click and it did work again;
Friday it was fine, pleasant and everything. And today it’s going to be
unbearable because there’s no air, eh … Twenty seven degrees, it’s a bit
much … Yes, I’m counting on you, eh. Thanks, Mr Raab, bye bye.
(Gardner-Chloros, 1991:98–99)
When the speaker was asked, after the recording, why she had switched so
much – she was entirely capable of speaking monolingually in either French or
Alsatian – the only answer she was able to provide was that her interlocutor had
been switching as well. Interruptions (marked with …) mark the points where
the other speaker was replying, but the whole extract was delivered at high
speed and the fifteen or so switches in no way interrupt the flow (note for
example the switch in L8 between the complementizer parce que and the
(abridged) pronoun es which follows). Since Mrs F’s switches do not noticeably
coincide with the other speaker’s contributions, there is no reason to think that
the actual points at which she switches are determined by her interlocutor –
though the overall decision to code-switch was allegedly connected with her
knowledge that he was a fellow code-switcher.
Mrs F is a typical Strasbourg code-switcher, but as one would expect in a town
of that size, there are many variations on the theme. CS varies from a limited
alternation which can be described in terms of “French plus Alsatian” (or viceversa) to a mixed code with some “third system” phenomena present. Example 2,
though within a private conversation, is typical of the switching which occurs as a
result of French dominance in official and administrative matters:
Example 2
mrs b: E promesse de vente isch unterschriwwe, nitt, awwer de contrat de vente nitt,
denn … der het’s kauft vor zehn Johr, am achzehnte janvier; s’il le revend
avant, hett’r e plus-value.
The exchange of contracts has been signed, you see, but the sales contract
hasn’t, because … he bought it ten years ago, on the eighteenth of January;
if he sells before (ten years are up), he’ll have to pay gains tax.
The use of French for the technical terms (promesse de vente, contrat de vente,
plus-value) and for the name of the month, ‘January’ – along with greetings,
numbers, time of day/days of the week/years, are typical features of Alsatian
switching. Here, they spill over, or “trigger”, in Clyne’s (1967) term, the phrase
s’il le revend avant. So even within such a short utterance, at least three types of
CS can be identified: switching for technical terms, for months/dates and owing
to triggering. A further motivation – accentuating the contrast between the if
clause and the rest of the sentence, i.e. a discourse-related reason, may well
account for the switch back to Alsatian in the last line.1 Although it is tempting
to say here that the “base” language, which provides most of the grammatical
“glue” is Alsatian, in many cases it can rapidly become impossible to say which
is the dominant language of the conversation, as below in Example 3.
The use of CS is not much constrained by topic in Strasbourg, but is associated
with informal contexts and a “chatty” register. In a comparison between family
and workplace settings, it was found to be more frequent, and more intensive, in
workplaces, between colleagues or peers than between family members, though it
is more frequent in families than between people who do not know each other. For
example, in an anonymous study carried out in three department stores, modelled
on Labov’s study of “r” in New York, salespeople code-switched far more among
themselves than when talking to customers (Gardner-Chloros, 1997). In families,
CS reflects inter-generational competence differences, i.e. the switches take place
more between speakers and less within their utterances. An example of this will be
found in Chapter 4.
Particularly intriguing are cases of CS with no apparent motivation. Such CS
is found in situations where both varieties are intrinsic to people’s identity –
what Myers-Scotton called “switching as an unmarked choice” (1983; 1988). In
such cases, in the words of Boeschoten (1998:21), CS begins to acquire
“language-like” properties. Speakers who never speak otherwise than in a
code-switching mode have been observed in many contexts (Swigart, 1991;
Meeuwis and Blommaert, 1998; McCormick, 2002).
According to Myers-Scotton’s theory (1993b), we are dealing here with Alsatian as a Matrix
Language and the conditional clause s’il le revend avant is an Embedded Language Island.
According to Muysken (2000), this represents Insertional CS. Grammatical interpretations of CS
are discussed in Chapter 5.
Social factors in code-switching
Example 3 is a case of such unmarked CS, in which the individual language
changes are not “indexed” with any particular significance. Three men in their
twenties, all fluent bilinguals, are chatting at work about their leisure occupations.
One of them, Gérard, starts excitedly describing a motorbike video game. The
immediacy of his action-packed description is reflected in his use of the present
tense, his short, staccato, sentences and his unconstrained CS.
Example 3
gérard: Wenn d’mit drüewwer witt, alors il cogne, alors la moto se renverse, puis il
faut la remettre sur pied. Moi, je suis arrivé à 80. S’isch e truc, wenn’s
e paarmol gemacht hesch, hesch’s hüsse, après il y a des difficultés,
kannsch e programme deux mache, noh muesch, pour pouvoir traverser,
isch dann d’distance entre le début et la fin de l’obstacle, un vorhär
hesch e so grosser Platz g’hett, disons.
If you want to get across with it, then he knocks, and then the motorbike
turns over and you have to right it again. I got up to 80. ‘Thing is, when
you’ve done it a couple of times, then there are difficulties, then you can
put on programme 2, then you must, in order to cross, there’s the distance
between the beginning and the end of the obstacle, and before that
you’ve had, say, this much space, let’s say.
(Gardner-Chloros, 1991:153)
Although the speaker starts in Alsatian, in the first part of the utterance, French
appears to be dominant, but Alsatian seems to take over more in the latter part.
This “turnover” in the dominant language is quite common, and it would be
difficult to assign a “base” or “Matrix” Language to the utterance as a whole
(Myers-Scotton, 1993a). Such examples challenge the usefulness of the Matrix
Language as a descriptive tool – an issue which is considered further in Chapter 5.
Signs of convergence or melding of the two varieties, going beyond the
phonological level and the borrowing of individual words or expressions, can
also be found in the Strasbourg context. They include the following:
(1) Expressions calqued from French into Alsatian and vice-versa:
e.g. es saat mir nix ‘it doesn’t appeal to me’ – modelled on the French
expression ça ne me dit rien, lit. ‘it doesn’t say anything to me’; achtung
mache ‘to pay attention’; French: faire attention, Alsatian: ufpasse,
German: aufpassen. Although the French expression has provided the
model, Alsatian word order (verb last) is used. This is an example of
“covert” borrowing, since the expression remains ostensibly in Alsatian.
(2) Composite word order, as in the utterance:
Toujours il était bon
French: il était toujours bon ‘it was always good’ v.
Alsatian: immer war er guet ‘always was it good’
The adverb placement at the beginning of the sentence is modelled on
Alsatian, but the subject–verb order is French.
(3) Bilingual verbs: like Standard German has done at various times in its
history, Alsatians freely borrow French verbs and integrate them thanks
to the ending -iere (or -iert for the past participle), e.g. camoufliere ‘to
camouflage’, engagiert ‘taken on, employed’, arrangiert ‘arranged’, confirmiere ‘to confirm’, forciert ‘forced’. Some of these have entered Alsatian
via German, but the process remains productive in the dialect, with examples such as demenagiere ‘to move house’, enregistriere ‘to record’ and
choisire ‘to choose’ being good examples of spontaneous CS creations
which can instantly acquire loan status owing to their being based on an
existing well-attested borrowing paradigm.
Alsace provides a useful example of historically rooted CS in a European
regional context. It illustrates how several trends and motivations can simultaneously spark off CS, leading to co-existing forms of it within the same
community. Despite its diversity, and despite the fact that in some cases it
signals the loss of the dialect, CS as a whole functions as a marker of being
Alsatian in the twenty-first century.
Greek Cypriots in London Background The Greek Cypriot community in London consists of
around 200,000 people (Christodoulou-Pipis, 1991), and is therefore comparably sized with the population of Strasbourg, though obviously more dispersed.
In other respects there are significant differences: the Cypriots represent an
immigrant community within multi-ethnic London, as opposed to an indigenous one surrounded by rural areas where related varieties are spoken.
Full-length studies of language contact in immigrant communities include
Auer (1984), Backus (1996), Clyne (1967; 1991b); Halmari (1997), Li Wei
(1994), Nortier (1990), Sebba (1993) and Zentella (1997). The range of linguistic phenomena in such contexts is as varied as in indigenous ones, though
Dabène and Moore (1995) point to a few characteristics commonly found in
migrant settings which are likely to affect the linguistic picture. First, the
circumstances of the migration have an immediate impact on the restructuring
of the different groups’ social networks; second, there may be a split between
the behaviour of those members of the community who develop close links with
the host society and those whose lives revolve around the home (for example,
housewives or the elderly), who may acquire only minimal skills in the host
language and/or take on a role as guardians of the home language. On the other
hand, adolescents are often strongly motivated to adopt the language habits of
the host community and this may lead to non-reciprocal language use within the
home. Finally, the mutual teaching of languages developed within families is
conducive to linguistic creativity and each family cell may become a “privileged setting for specific linguistic behaviour and interactional patterns” (p. 30).
Social factors in code-switching
Differences between the generations in the London Greek Cypriot community are heightened by the fact that different waves of immigration occurred in
different circumstances. Immigration started in the 1930s, for economic reasons. Most migrants of that period were peasants with little schooling, who
moved to England because they did not have enough land to live off at home.
Some who are still alive now can be found in north London community centres
and lunch clubs, sitting round in communal rooms in single sex groups as if in a
village square in Cyprus, the women all in black and wearing black headscarves, the men playing backgammon or watching TV and drinking coffee.
Many of them never learned English fluently, but they borrowed from
English a large number of words which they adapted to the Cypriot dialect,
and which are now considered the archetypal marks of London Greek Cypriot
Dialect or GCD (see below). They worked in the catering or hotel trade, clothing
or shoe manufacturing or hairdressing and grocery shops. Often wives and
families joined the breadwinner after he had been established for a few years
(Anthias, 1992). Many women remained in the home, taking in “part-work”,
which they did at a sewing machine in the kitchen. Most settled in the Borough
of Haringey, around Green Lanes, where there are still a succession of Cypriot
shops and offices (doctors, lawyers, travel agencies, etc.). Children were sent to
Saturday schools, of which there are around fifty. There they were taught Greek
by teachers sent over by the Greek government, who made no concession to the
fact that their pupils spoke a dialect at home which differs in significant respects
from Standard Greek.
The largest migratory wave, in the 1950s and 1960s, was again dictated by
economic factors. Finally, after the Turkish invasion of Cyprus in 1974, a
number of refugees from all walks of life came over to Britain to make a fresh
start. The children of this latter group, having often started school in Cyprus,
stand out for their proficiency in Greek; many take GCSE Greek. Each of these
waves now comprises several generations, and the youngest in the community
are on the whole heavily anglicized. Nevertheless, the rate of intermarriage with
the local non-Cypriot population has been relatively low. Many younger people,
when interviewed, said they would prefer to marry another member of the
community rather than either an English person or a Cypriot from Cyprus
(Gardner-Chloros, McEntee-Atalianis and Finnis, 2005). Types of code-switching English words, principally nouns, were adopted and morphologically/phonologically adapted into the London GCD at an
early stage in the migration history, and have become part of the community
repertoire – known as BBC (‘British-born Cypriot’) Grenglish. Some, like
marketa ‘market’ have displaced native equivalents, others, like the expression
for ‘fish and chip shop’, have been coined to cover concepts which would not
have been relevant in Cyprus.
Example 4
BBC Grenglish
fish and chip shop
[no equivalent]
[no equivalent]
Such items are loan words in that they are used community-wide in preference
to the “native” equivalent, as well as being morphologically and phonologically
adapted to GCD. The younger and more balanced bilinguals use them alongside
unadapted code-switches, i.e. English nouns or compounds in a GCD framework (e.g. situation, GP, flat, second cousins, Mediterranean, privacy, factory,
nine-to-five jobs, dinner dance, receptionist, very tight community, etc.). As was
the case with French–Alsatian switching, there are also switched adjectives,
adverbs, conjunctions and, less commonly, other parts of speech. Appel and
Muysken (1987:170–171) refer to a “switchability hierarchy”, in which nouns
tend to be the most frequently switched elements owing to the referential weight
which they carry.
Although nouns may be switched more frequently, a frequent type of CS
involves the formation of new verbs. A common way of doing this in Alsatian,
as we saw, was by using the productive verb-ending -iere. In GCD–English, the
Greek verb ending -aro is added to various English verbs (e.g. kanselaro ‘to
cancel’, tʃekaro ‘to check’, muvaro ‘to move’, fixaro ‘to fix’). A further technique
involves formations using a verb meaning ‘do/make’ (in this case the Greek
Cypriot kamno) in combination with a verb, noun or adjective from the other
language, to make a new compound verb. This construction has been found to be
prevalent in several diasporic Greek–English contexts: Montreal (Maniakas,
1991), Chicago (Seaman, 1972) and in Victoria, Australia (Tamis, 1986). It is
of particular interest in CS as it constitutes a widespread “intermediate” or
innovatory process found not only in Greek–English but in a huge number of
CS contexts (see 2.3.2).
Example 5
[Mixed compound verb formations in London Cypriot Dialect]
kamno use
to use
kamno respect
to respect
Social factors in code-switching
kamno developed
to develop
kamno spelling
to spell
kamno improve
to improve
[Occasionally the verb eχo ‘to have’ is also used in such formations, as in:]
Ksero oti eχis very busy I know you are [lit. ‘have’] very busy
The fact that the speaker uses ‘to have’ rather than ‘to be’, which would be
closer to the relevant expression in either language, suggests that we are dealing
with a new verbal formation rather than straightforward CS. Perhaps there is
underlying interference from the expression ‘to have a lot of work’, which is the
same in Greek. Perhaps also the speaker cannot recall, or does not know, the
Greek for ‘busy’, aposχolimenos, which is long and erudite compared with
busy. The following examples present similar questions:
Example 6
ekamna upset
I was upset
Here the past tense of kamno is used instead of the past tense of ‘to be’ (lit. ‘I made
Example 7
I agli
tus ksenus
the English NEG do
except (the) foreigners
the English do not make exceptions for foreigners
Here kamno + except departs from both English and Greek models. The
English would be make an exception for and the Greek would require either a
monomorphemic verb, or, more colloquially, an expression exactly equivalent
to the English one. Grammatically deviant examples Such examples show that there
is a grey area where borrowing, code-switching, convergence and innovation
cannot easily be distinguished. The tidiness of the switchability hierarchy breaks
down as soon as the switched element is a complex, rather than a simple, noun
phrase, e.g.:
Example 8
to mono pu mu aresi δamesa ine oti ise very tight community
the only thing which I like here is that you are very tight community
The compulsory article required in English before the switched phrase very tight
community is missing, which violates the equivalence constraint (see Chapter 5).
Example 9
bori na δiavazi ke na γrafi ala oχi ke a hundred percent
she can read and write but not actually [lit. and] a hundred percent
This is another violation, as the Greek for ‘a hundred percent’ (ekato tis ekato)
does not take an article. Morphosyntactically deviant structures also arise when
the obligatory subject is omitted, in utterances such as ‘is’ for ‘it is’ below:
Example 10
ama thelis na χrisimopiisis ta trapezomandila tus, ta pramata tus, is not inclusive
if you want to use their tablecloths, their things, is not inclusive
Example 11
kseri ime kipreos tshe nomizo oti suspect you an men tu miliso ellinika
he knows I am a Cypriot and I think (he) suspect(s) you if I (actually) speak Greek to
The verb suspect is a “bare form” with no ending. It also lacks a pronoun subject
from the point of view of English and the VO order is incorrect from the point of
view of Greek.
Switches are also found at locations where adjective placement rules conflict,
as in this unpublished example:
Example 12
irthe δaskala private
came teacher private
a private teacher came
This example is of interest from the point of view of constraints and is
commented on in section 5.3.1. Symbolic duality Below, as in the case of Alsatian, a longer extract
from a typical conversation has been selected. The interviewer was a Greek
Cypriot from Cyprus, and the respondent is a second-generation member of the
London community. The passage is revealing in terms of content as well as CS.
Maro is linguistically typical of teenagers and young people, who are not
generally fluent bilinguals, and many of the switches appear to be due to a
lack of fluency in the GCD.
Example 13
[Conversion with Maro, age 20, social worker]
1 o: Se pio meros menete tora?
where do you live now?
2 m: Palmers Green
3 o: Thelis na pas piso stin kipro i thelis na minis eδo mesa?
Do you want to go back [sic] to Cyprus or do you want to stay here?
4 m: δen thelo na mino eδo ala pros to paron –
I don’t want to stay here but for the time being –
Social factors in code-switching
6 o:
7 m:
9 o:
10 m:
12 o:
13 m:
um, I don’t want to go back to Cyprus but I might go to another country but it
won’t be Cyprus
because persi to kalokeri.. um …δen mu aresane ta … ta provlimata
last summer
I didn’t like the … the problems
you know I didn’t like- I didn’t like the way – I don’t want to sound like an
outright feminist but I don’t – I don’t like the way they treat the women …
like my cousins right – [to another interlocutor who says something
inaudible in the background] – YOU INTIMIDATED ME – they’ll be
sitting there and their husbands will come in and go: kori sofia, piene
ferta ruχa mu ia na kano banio
Sophia my girl, go and fetch my clothes so that I can have a swim
and I mean things like that, know what I mean. I can’t … δen boro na sinithiso …
I can’t get used to …
niothis kseni otan pas kato?
do you feel foreign when you go down? [i.e. to Cyprus]
ne niothoyes I feelone of the things I like when I go to Cyprus is that the language is not alien to
you, people are speaking the same language that you were brought up with
so that’s – you know you get a sort of warm feeling from that, but still I do
feel like a foreigner over there
uh..δilaδi vriskis provlima na milisis kato kipriaka?
you mean you find it difficult to speak Cypriot when you’re there?
oχi δen..δen vrisko provlima-ine oti
no I don’t..I don’t find it difficult-it’s that
you know ine orea otan pas kapu ke milun ti γlosa pu katalavis2 you know
it’s nice when you go somewhere and they speak the language which you
In L5, 7 and 10, Maro breaks off speaking GCD to explain what she means, or to
play for time (you know) in English. She is ill at ease because of her halting –
though by no means elementary – Greek – but perhaps also because she is being
questioned about her identity and life plans by an interviewer from Cyprus, i.e.
effectively a foreigner, in front of giggling friends who, at one point (L7), put
her off her stride by saying something in the background to which she reacts
strongly. She is also being questioned about speaking GCD in GCD, in which
she is not fully fluent, in the context of a rather tricky conversation about
feelings and allegiances. In L12 there is an indication that the interviewer
herself has not properly understood Maro’s comments, and that is why Maro
tries to rephrase them in GCD in L14.
Incorrect ending. The correct form would be katalavenis. Maro makes various “mistakes” in her
Maro claims that she does not have a problem with speaking GCD (L14)
and that it gives her a warm feeling to hear the language spoken in Cyprus.
She is at her most fluent and confident when she imitates her cousins’ macho
remarks in Cyprus, showing that she identifies with the values of British youth –
equality for women. In this she identifies with the values of her country of
residence, the UK. Older members of the immigrant community would tend to
be more conservative than the society where they live, hankering after an
outdated image of their country of origin. Other girls of a similar age and
background said that they would neither want to marry a Cypriot from
Cyprus, nor an English person, but only a Cypriot from Britain.
The conversation is revealing at various levels, not least the psychological.
Maro’s attitude to GCD is like that of an adolescent towards their family – torn
between feelings of belonging and the desire at the same time to distance
herself, to express a separate identity. The “dual voicing” which CS allows
symbolizes this split (see Chapter 4). In conversations with peer-group members, Greek Cypriots of Maro’s generation make use of CS for a variety of
expressive and identity-creating/reinforcing purposes (Gardner-Chloros and
Finnis, 2004).
These two case-studies, the Alsatian and the Cypriot, show how the study of
CS in any given community brings in factors at many different levels and, of
necessity, involves some ethnographic knowledge and engagement. In both
these cases, for example, there are significant differences between the generations in terms of their motives for CS; these difference stem from the speakers’
personal and collective history. Considerable caution should be exercised in
relation to analyses, be they grammatical or psycholinguistic, which rely on decontextualized data.
Macro-linguistic approaches
Sociolinguistics covers a wide range of issues, from language policy and the
description of “linguistic markets”, via the different linguistic behaviour of
women v. men, middle class v. working class, and other social groups, right
down to conversations between individuals. Similarly, CS can be studied at the
level of whole societies where multilingualism is prevalent (e.g. India), but is
also telling at inter-individual and idiolectal level, being closely bound up with
the ways in which conversation is structured.
Gumperz and Hernandez wrote that “CS occurs whenever minority language
groups come into close contact with majority language groups under conditions
of rapid social change” (1969). However, the majority of studies have placed the
emphasis on explaining CS from a microlinguistic perspective rather than on
showing how it relates to broader aspects of the societies where it occurs. In the
next section a few of the broader-based studies are described.
Social factors in code-switching
Heller’s (1988a) volume Codeswitching: anthropological and sociolinguistic
perspectives contains various papers in which CS is described in terms of its
connection with characteristics of the communities where it occurs: Heller’s own
paper (1988b) shows how CS can be used to manage and avoid conflict when
different varieties are associated with different roles in a society. She gives
examples from a Montreal company and a school in Toronto to show how CS
allows people to gain access to different roles or “voices” by switching from
French to English or vice-versa, and thereby exploit various ambiguities inherent
in the situation. Woolard (1988) describes a comedian, Eugenio’s, use of Catalan–
Castilian CS in Barcelona, not as a test of in-group membership but rather as a
way of addressing two audiences at once and thereby levelling, rather than
maintaining, the boundary between them. McConvell (1988) describes switching
between dialects of an aboriginal language, Gurindji, and English in terms of the
inter-related social “arenas” where these are used, as Denison had done earlier
(1972) for Sauris, a German-dialect speaking enclave in the Friulian Alps.
Gal (1988) points out that CS often involves one state-supported and one
stigmatized minority language. Vernacular linguistic forms continue to be used
because they represent a form of resistance to domination, so such patterns of
use do not simply reflect the socio-political situation, they help to shape it. The
latter point is an important one: several others have also pointed out that
traditional sociolinguistics tends to present the stratification which it portrays
in society (e.g. class or gender based) as if it were the result of a consensus, and
thereby to gloss over the fact that the observable differences may in fact embody
conflict or dissatisfaction (Williams, 1992; Cameron, 1992; Pujolar, 2001). As
Cameron puts it, “The language reflects society account implies that social
structures somehow exist before language, which simply ‘reflects’ or
‘expresses’ the more fundamental categories of the social” (1997:57).
Diglossia, marked choices and networks
Ferguson’s description of certain linguistic situations as “diglossic” (1959; reprinted in Li Wei, 2000), continues to form a useful basis for discussing bilingual
situations. This is not because the diglossic communities described by Ferguson
are unchanged – the description was not totally accurate even when first written –
but because these proposals focused attention on the functional differences
between different varieties of the same language and provided a set of structural
parameters which allowed one situation to be compared with another. Language
use in bidialectal situations – the model was subsequently extended by others to
bilingual ones – is described in terms of complementary domains3 of usage, of the
Fishman defines domains in terms of “institutional contexts and their congruent behavioral
co-occurrences”, e.g. family, employment (1972:441).
varieties’ relative prestige, their role in official life, religion, education and
literature. The schema was the subject of significant amendments by Fishman
(1965; 1967; also reprinted in Li Wei, 2000). Breitborde (1982) subsequently
pointed to some difficulties with connecting the abstract notion of domain with its
impact in actual interactions: the features which make up a domain are rarely a
perfect fit, so in each case some aspects are likely to be more significant than
others. The concept of diglossia was discussed in relation to CS by Myers-Scotton
(1986), via the notion of marked and unmarked choices, mentioned above and
further described in Chapter 4. Markedness Theory was developed in the context
of explaining the socio-psychological motivations for CS, using data collected in
various settings in Africa, Kenya in particular (Myers-Scotton, 1993a).
Li Wei, Milroy and Pong Sin Ching (2000) proposed social networks as an
alternative means of relating CS and the language choices of individuals to the
broader social, economic and political context. They claim that a social – as
opposed to sociolinguistic – theory which associates network patterns with the
sub-groups which emerge from political, social and economic processes,
remains to be developed. Højrup’s (1983) division of the population into subgroups described in terms of different “life-modes” provides one possibility. Li
Wei et al. found that these life-modes corresponded well with the linguistic
behaviour of members of different types of network in their study of the
Tyneside Chinese.
The Gumperz tradition
John Gumperz, whose early work on CS put it on the sociolinguistic map,
investigated it in contexts ranging from Delhi to Norway, from the point of view
of its historical genesis, its linguistic consequences, its significance for speakers
and its conversational functions. Some of this work has already been discussed
in Chapter 2. Here, the discussion concentrates on two aspects of his analysis
which continue to be influential: the notions of we-code v. they-code, and the
distinction between situational and conversational CS. Much of Gumperz’s
earlier work on CS, originally published in less accessible sources, was recapitulated in Discourse Strategies (1982), so for the sake of convenience most of
the references here are to that volume.
We-codes and they-codes
As a direct consequence of diglossia, Gumperz (1982) suggested that the
ethnically specific, minority language comes to be regarded as a “we-code”
and to be associated with in-group and informal activities, whereas the majority
language serves as the “they-code”, and is associated with more formal, outgroup relations. However, he emphasized that the relationship between the
Social factors in code-switching
occurrence of a particular set of linguistic forms and the non-linguistic context is
indirect, and that there are only very few situations where one code exclusively
is appropriate. “Elsewhere a variety of options occur, and as with conversations
in general, interpretation of messages is in large part a matter of discourse
context, social presuppositions and speakers’ background knowledge” (p. 66).
In CS, the we-code and the they-code are often used within the same conversation, as in Example 14. A Punjabi–English bilingual, Umi, talks to a friend
about the likely loss of Punjabi culture in Britain
Example 14
umi: culture tha
aapna …..rena tha
culture [tha = stress marker] our …… stay [stress marker] is-not
we know it, we know it, we know it’s coming
Our culture is not going to last, we know it, we know it, we know it’s coming
(Gardner-Chloros, Charles and Cheshire, 2000: 1322)
The threat to Punjabi culture is poignantly embodied in the switch from the
we-code to the they-code half way through the sentence, and by the use of the
English word for ‘culture’.
From an early stage, variations on the classic we-code/they-code dichotomy
have been reported. Singh (1983) wrote that, although the minority language is
usually the we-code, this is not always the case. In India, for example, speakers
with social aspirations may use English as their we-code and Hindi with ironic
intent, to show themselves to be a different kind of minority, whose apartness
is based on privilege. Sebba and Wootton (1998) also state that even where
there are two or three distinct codes available, a multiplicity of social identities
may be evoked and manipulated through them, and the relationship between
code and identity is far from being one-to-one. They illustrate the point by
showing unexpected configurations of we- and they-codes in various contexts:
Cantonese is the we-code in Hong Kong classrooms, where English is learned
as an L2, but cannot be equated to an insider code as Cantonese is the majority
language. For British-born Caribbeans, London English and London Jamaican
are both we-codes, since it is the ability to use both which characterizes the
“Black British” speaker.
The we/they-code distinction breaks down in situations such as that described
in Meeuwis and Blommaert (1998). In the Zairian community in Belgium,
CS can be a variety in its own right, with the same functions and effects as
those usually attributed to “languages”. In communities where this is the case,
speakers vary in the extent to which they are able to speak the two varieties
monolingually. All the national languages of Zaire are spoken as CS varieties
peppered with French (French being the official language in Zaire).
Lingala–French and Swahili–French code-switched varieties (Lingala and
Swahili being the two most widespread national languages), have their own
range of social, stylistic and register-related variation. A similar situation is
reported in Swigart (1991) with respect to the CS variety known as Urban Wolof
in Dakar. Such cases point to the dangers of viewing CS from a monolingual
reference point in which meaning is seen as being negotiated through the
interplay of two differentially marked “languages”. Beyond this, there are
cases where the we-code/they-code distinction completely fails to account for
the variation and CS which are observed. Instead, the contrast between the two
varieties is used to bring about “local” meanings in a variety of ways, only some
of which make use of the associations of the two languages (see Chapter 4).
Indeed the adoption of CS may in itself be an “act of identity”, a fact which
we see clearly in the case of “crossing” (Rampton, 1995). Rampton describes
adolescents in Britain using features of Punjabi and Creole in order to create a
trans-racial “common ground”. By contrast with other types of CS, crossing
“focuses on code-alternation by people who are not accepted members of the
group associated with the second language they employ. It is concerned with
switching into languages that are not generally thought to belong to you”
(p. 280; see also Hewitt, 1986; Jǿrgensen, 2005). Franceschini (1998) gives a
bilingual example:
Example 15
In a fashion house in Zurich, I am served by a ca. eighteen-year-old shop assistant in
Swiss-German. After about ten minutes, a group of young men, obviously friends of the
shop assistant, enter the shop. All of them use the common Swiss-German/Italian CS
style, which is certainly not surprising. There is nothing unusual about the scene. The
group seems to me to be one of many second-generation immigrant peer-groups. In order
to exchange my purchase, I go to the same fashion house the following day. I am now
served by the owner of the shop, a ca. forty-year-old Italian. In the course of our
conversation, I am told that the shop assistant I overheard the previous day is not a
second-generation Italian immigrant at all but a Swiss-German. She grew up in a
linguistically strongly mixed area of the town and has had Italian friends since her school
years. (Franceschini, 1998:56–57)
The young shop assistant code-switches, not out of linguistic necessity, but in
order to identify herself with a particular peer-group. However, CS due to
necessity and CS as the product of choice, are not always easy to separate.
Many instances of CS are combinations of the two, or somewhere on the border
between the two. Auer (2005) shows that it is not always easy in practice to
disentangle discourse-related CS (see Chapter 4) from such displays of identity.
Situational and conversational code-switching
Equally influential with the we-code/they-code distinction was Gumperz’s subdivision of CS into situational and conversational types. Situational CS occurs
Social factors in code-switching
when distinct varieties are associated with changes in interlocutor, context or
topic, and is therefore a direct consequence of a diglossic distribution of the
varieties. Conversational CS occurs when there are changes in variety without
any such “external” prompting. Such switching is further termed metaphorical
when the purpose of introducing a particular variety into the conversation is to
evoke the connotations, the metaphorical “world” of that variety. Blom and
Gumperz (1972) gave the example of two villagers in a social security office in
the Norwegian village of Hemnesberget, switching from Standard Norwegian
to discuss business, to the local dialect to discuss family and village matters.
Although this type of switch, and the compelling motivation for it, are familiar
to anyone who has observed CS in this type of minority situation, it was asserted
by Maehlum (1990) that the dialect and standard varieties taken as a prototype by
Gumperz were in fact “idealized entities” which in practice are subject to
interference at different linguistic levels: “Most probably, the switching strategies
which Blom and Gumperz recorded in Hemnes actually represent some form of
variant switching whereby, in certain contexts, single words, (idiomatic) expressions and grammatical forms from the standard are introduced into otherwise
dialectal utterances” (1990:758). Maehlum claims that the misapprehension is
due to the researchers’ insufficient knowledge of the ins and outs of the dialectal
situation in that area. Gumperz himself remarked that recordings of informal
conversations in the same town, which speakers claimed were conducted entirely
in the local dialect, “revealed frequent conversational switching into standard
Norwegian” (1982:62). Along with classic diglossia, situational CS appears to be
a somewhat idealised notion, rarely found in practice.
Comparisons between and within communities
As we saw in Chapter 1, one of the challenges posed by CS is to explain the
variation within it, or, viewed another way, to decide how broadly it should be
defined. It has been defined here as inclusively as possible, because, in the
present state of knowledge, it has not been demonstrated that the differences
between CS and other language contact phenomena are categorical differences
as opposed to differences of degree. CS merges with lexical borrowing at one
end of the scale, one of the most “minimal” manifestations of contact, and with
convergence/interference/code-mixing at the other end, which can be seen as
the last step before total fusion. If the process of language contact always started
and ended in the same way and always proceeded along a similar path, it would
be easier to divide it into distinct phases. Instead, our task is a messier one – to
try and apprehend the variations involved and to tie them in with the factors
which may help explain this variety. Variation in CS can, however, be divided
for practical purposes into variation between communities and variation within
communities. Variation within communities can be considered from several
points of view, including age, gender and network, some of which are discussed
elsewhere (see above and Chapter 4). Some of the studies of variation between
communities are discussed below.
Variation between communities
Making systematic comparisons between CS in different language combinations and different contexts is the best way to elucidate the contribution of
typological factors on the one hand, and sociolinguistic ones on the other, to the
patterns of CS in different communities. So far, only a few such comparisons are
available. On the whole, researchers base their discussions of CS on their own
data, collected in a single community, and do not have access to comparable
data-sets from other communities. The LIPPS Project, mentioned in Chapter 1,
has as its purpose to set up a database of CS texts coded according to a common
protocol and thus facilitating such comparisons (LIPPS Group, 2000) (see
Appendix). Meanwhile, some existing comparisons between communities or
sub-groups are discussed below. Treffers-Daller (1994;1999), discussed in
Chapter 2, Cheshire and Gardner-Chloros (1998), discussed in Chapter 4 and
Muysken (2000), discussed in Chapter 5, also employ a comparative approach.
Comparisons between communities McClure and McClure (1988) McClure and McClure (1988) set out
to take a broader perspective than in much CS research by describing a multilingual Saxon community in Romania in terms of the macrolinguistic relationships between the groups. The Saxon and Romanian communities are quite
separate, but unlike other minority groups, the Saxons do not occupy a subordinate position vis à vis the majority. Consequently, their CS is more limited
in type than that described elsewhere. Situational switching, mainly dictated by
changes in participant, is dominant over the conversational variety. Where the
latter does occur, its main function is to highlight quotations.
McClure (1998:141) compares written CS between English and the national
language in Mexico, Spain and Bulgaria. The characteristic type of CS encountered in each of these countries reflects the functions of, and attitudes towards,
English. In Mexico and Spain, English is widely known, and is used in the press
in various expressions denoting concepts expressed more economically in
English or using “English” concepts (e.g. Latin lover). But in Mexico, which
shares a frontier with an English-speaking country and resents the latter’s
economic and cultural hegemony, CS is functionally richer than in the other
two settings. It is used in ironic contexts to reflect a certain rejection of the
US culture, as for example in the use of by the way in this quotation from the
Mexican press:
Social factors in code-switching
Example 16
La hipocresia norteamericana no estriba tanto en los lamentos exagerados por la muerte
de un agente de la DEA, y en la indiferencia o incluso el desprecio ante la muerte de
decenas de agentes mexicanos (o, by the way, de miles de civiles panameños).
The North-American hypocrisy does not rest so much on the exaggerated laments over
the death of an agent of the DEA, and on the indifference or even the scorn with respect to
the death of tens of Mexican agents (or, by the way, thousands of Panamanian civilians).
(Proceso, 15 January 1990).
By contrast, in Bulgaria, English has increasingly been used, since the fall of
the Communist regime, as a symbol of the West, a cultural and economic world
to which many Bulgarians aspire. English is not yet sufficiently well known for
more subtle uses of CS, but is widely present in advertizing and in other
documents such as the Yellow Pages for Sofia. Poplack (1988) Poplack (1988) made a three-way comparison
between data collected in the Puerto-Rican community in New York (Poplack,
1980) and a later data-set from five neighbourhoods within the Ottawa–Hull
community in Canada, which is divided by a river which constitutes both a
geographic and a linguistic border (not in the dialectician’s sense of “isogloss”,
but in a sociolinguistic sense). On the Quebec side (Hull), French is the official (and
majority) language, and on the Ontario side (Ottawa), it is a minority language. The
comparison is of particular interest as any differences between the communities
must presumably be attributed to the different status of French in the two communities. The method of data collection and the definition of what constituted CS was
the same in both cases. For the purposes of this study, Poplack considered as CS the
use of English material in the context of French conversations (i.e. in practice she
operated with the notion of French being the base language). In keeping with her
view as to the demarcation line between CS and borrowing (Poplack and Sankoff,
1984), she did not count single English words which were morphologically or
phonologically integrated with French as CS.
The most striking finding was that in the Ottawa communities, where French
is a minority language, CS was “three to four times as frequent” as in Hull, i.e.
the stronger influence of English in the environment was directly reflected in the
amount of CS (1988:226). The same switch-types were found in both communities, but the distribution of the four main types was radically different.
Tabel 3.1 shows that in both communities the commonest switches were mot
juste switches, switches for metalinguistic comments, switches where the
English intervention is “flagged”, as in:
Example 17
Excuse mon anglais, mais les odds sont là
Excuse my English, but the odds are there
Table 3.1. Functions of code-switching in five Ottawa–Hull neighbourhoods
(a) Four sample members whose use of English greatly exceeded that of the other informants and
whose status as French L1 speakers is not clear, were excluded from this study.
(b) Asterisks indicate that the effect is essentially due to that number of individuals.
(c) Percentages may not add up to 100 due to rounding.
and switches in the context of explaining/translating. This points to a fairly selfconscious use of English in both cases, with switches in Quebec being largely
restricted to metalinguistic commentary, which, as Poplack points out, show the
speaker’s full awareness of using English, e.g.:
Example 18
Je m’adresse en français, pis s’il dit “I’m sorry”, ben là je recommence en anglais
I begin in French and if he says, “I’m sorry”, well then I start over in English.
Poplack comments that this reflects the fact that in Hull, people believe that
good French must of necessity exclude Anglicisms. Reasons underlying the differences The comparison with Puerto
Ricans in New York is less direct. There is a wide range of differences between
the two situations, such as the fact that the Puerto Ricans are of immigrant
origin, which could account for the differences in the prevalent types of
CS. We are also dealing with another language combination, but as Poplack
Social factors in code-switching
points out, the linguistic distance between English and Spanish is not much more
remarkable than that between English and French. It is probably more significant
that different data collection techniques were used, the Puerto Ricans being
studied through participant observation, whereas the Canadian studies were
conducted by means of interviews with out-group interviewers. The latter technique might mask the extent of CS and give rise to heightened purism on the
speakers’ behalf. Regardless of this, Poplack describes switching found in New
York as fluent and varied, with many unflagged switches, as opposed to the
limited, and more stilted, CS found in the two Canadian contexts. She ascribes
this mainly to the fact that for the Puerto Ricans both languages are an intrinsic
part of their identity and of their communicative practices. The cohabitation of the
two varieties within CS is a natural consequence of this integrated duality.
CS therefore arises, in different forms, in a wide variety of sociolinguistic
circumstances. There are also, more unusually, communities which appear to
shun it, as in the case described by Sella-Mazi (1997). This is the Moslem,
Turkish-speaking community in Thrace (Greece), who were afforded a special
status and protection of their linguistic rights under the Treaty of Lausanne
(1923). Although the younger members of this community are perfectly fluent
in Greek, they are described as avoiding CS, owing to a high level of awareness
of the need to protect their language and culture from Greek influence. A second
reason given is that the two languages are of widely differing importance in
terms of speaker attitudes.
Broadly “sociolinguistic” approaches to CS are extremely varied and cover
multiple levels of engagement with plurilingual data, from the societal to the
intra-individual. Dividing these approaches up is a partly arbitrary exercise,
since the societal level and the individual are in constant dynamic interaction.
The issue of gender, for example, could have been treated here with other major
sociolinguistic variables; instead, it is discussed in Chapter 4, as it appears to be
bound up with more detailed conversational factors rather than being a broad
differentiator within CS behaviour.
CS should be seen as a technique which has clear analogies in the monolingual sphere. Theories linking the social phenomenon of register and style
variation to individual performance are highly relevant to CS (see for example
the papers in Eckert and Rickford, 2001). Barrett (1998), for example, has
illustrated style alternation, largely in phonological terms, amongst African
American drag queens, identifying three basic styles (AAVE,4 gay male style,
African American Vernacular English.
and a style based on stereotypes of white women’s speech) and showing how
the subjects’ performances are “tuned” to highlight the audience’s assumptions
about sex, class and ethnicity (see also De Bose, 1992; Mishoe, 1998).
The concept of “audience design” developed by Bell (1984; 2001) and
Coupland (1985), can help explain many cases of CS, such as Example 1
above. Mrs F herself explained her abundant CS on the telephone in terms of
reciprocation of her interlocutor’s style, and furthermore, during the course of a
working day, she switched between several different styles of CS – and monolingual speech – depending principally on her interlocutor (Gardner-Chloros,
1991:92–94). The effect of audience design/accommodation on CS is well
illustrated in Wei Zhang (2005), where callers to a radio phone-in programme
in a bidialectal area of China are addressed in Cantonese by the host unless they
themselves reply in Putonghua, in which case the host switches to match. It is to
be hoped that in future, there will be more studies systematically comparing CS
and dialect – as well as style/register-shifting.
More recently, the linguistic styles adopted by individuals have become an
important focus of interest in sociolinguistics (Bell, 2001; Coupland, 2001;
2007; Eckert and Rickford, 2001). Broad, quantitative approaches which
obscure the differences between individuals are being put into perspective by
approaches such as that of Eckert (2000) based on the notion of “community of
practice”. This notion too could prove extremely useful in relation to CS, which
is, after all, no more than the bilingual manifestation of universal discourse
Code-switching in conversation
In Chapter 3, CS was considered as a sociolinguistic phenomenon – a linguistic
product of language contact, determined in various ways by the social circumstances in which it occurs. One of the principal challenges in CS research is to
determine to what extent the social circumstances affect the form which CS
takes in any given case. As we will see in Chapter 5, there is also a great deal of
research which emphasizes the linguistic and typological factors which shape
CS, and it is often considered that the CS patterns found in any given context
represent a choice among grammatical options, which are themselves defined
by the contributing languages (Halmari, 1997; Muysken, 2000).
Although there is evidence that typological factors influence the type of CS
which is found at a grammatical level (Muysken, 2000), there is no hard and fast
evidence that CS is constrained in any absolute way by the characteristics of the
languages involved. First, proving that it was would involve proving a negative –
that some kinds of CS could never occur. Second, this would imply that speakers
cannot break, or rewrite “rules”. But we should consider carefully what we mean
by “rules” when talking about language. As Le Page (1997) has observed, “A
language is best thought of as a game in which all the speakers can covertly
propose and try out rules, and all the listeners are umpires” (p. 32). Since we know
that certain kinds of CS are often found within certain language combinations,
and that certain kinds can be expected in certain social circumstances, the
challenge is to find the limits of these influences – no more, no less. This can
best be achieved by carrying out systematic comparisons between CS involving
the same language-pairs in different sociolinguistic contexts, and between
different language-pairs in similar ones.
In this chapter we consider three further sociolinguistic determinants of CS:
conversational/pragmatic motivations; social psychological influences, including Attitudes and Communication Accommodation Theory (CAT); and gender
(with some reference to Politeness Theory). The first of these has been written
about extensively in relation to CS; the other two have lagged behind, although
they are the subject of massive research efforts in their own right. It goes without
saying that each instance of CS, being a highly complex form of speech
behaviour, requires a multilevel explanation, so all these aspects, and more,
may be relevant simultaneously. The separate presentation of these factors here
therefore inevitably involves some oversimplification.
Conversational/Pragmatic motivations
Myers-Scotton (1993a:49) drew a distinction between the “allocational” paradigm, in which social structure determines language behaviour, and the “interactional” one, in which individuals make “rational choices” to achieve their
goals (p. 49). Milroy and Gordon (2003: Chapter 8) similarly contrast pragmatic
uses of CS which exploit the symbolism or connotations of each of the codes,
and those which purely exploit the contrast which the two varieties provide,
regardless of their connotations. They emphasize the need to pay attention to
both aspects in order to achieve a full understanding of CS.
Using the external symbolism of the two codes
Gumperz (1982) provided examples of both uses of CS. Using the distinction,
outlined in Chapter 3, between the we-code and the they-code, he described the
conversational functions of CS with examples from Slovenian–German,
Hindi–English and Spanish–English CS in the USA. He conceded that the
range of interpretations that results is much greater than one would expect
from describing the language usage in terms of the simple “we” and “they”
dichotomy. In the conversation below, Gumperz describes the speaker as alternating between talking about her problem in English and acting out her problem
through Spanish: the code contrast “symbolize(s) varying degrees of speaker
involvement in the message”:
Example 1
[A Chicano professional talks about her attempt to cut down on smoking]
They tell me “How did you quit Mary?” I don’t quit. I … I just stopped. I mean it
wasn’t an effort that I made que voy a dejar de fumar por que me hace daño o (that I’m
going to stop smoking because it’s harmful to me or) this or that uh-uh. It’s just that I used
to pull butts out of the waste paper basket yeah. I just used to go look in the … se me
acababan los cigarros en la noche (if my cigarettes would run out on me at night). I’d
get desperate y ahí voy al basarero a buscar, a sacar (and then I go to the wastebasket
to look for some, to get some) you know. (Gumperz, 1982:81)
Another example shows German–Slovenian bilingual Austrian farmers
discussing the origin of a certain type of wheat, and switching from Slovenian
to German in order to give countering statements more authority. This is typical
in that the majority language is often linked to objectivization or depersonalization of the statement.
Code-switching in conversation
The symbolic associations of the two codes are also often exploited in advertisements, as we have seen in the case of McDonald’s of (Box 1, p. 6). All over the
world, adverts make use of English words, phrases and slogans to convey images of
modernity, Westernization, etc. (Bhatia, 1992; Martin, 2002; Takashi, 1990;
Cheshire and Moser, 1994). Often the contrast between the local script and the
Roman characters used for the English insertions provides an iconic equivalent to
the different accents or voices which characterize oral CS – as in the case of the
Coca-Cola logo, embedded the world over in hundreds of languages on thousands
of hoardings. The fact that code-switching sells products is attested by the fact that
some of the biggest multinational companies use it. In Example 2, the ‘foreign’
flavour is introduced by the use of a different script.
Code-switching as a discourse-structuring device
Goffmann’s concept of footing, developed with reference to monolingual
speech, is highly relevant in CS: “A change in footing implies a change in the
alignment we take up to ourselves and others present as expressed in the way we
manage the production or reception of an utterance” (1979:5). Gumperz
referred to CS as a “contextualization cue”, that is a “verbal or nonverbal cue
that provides an interpretive framework for the referential content of a message”
(1982:131). His list of conversational functions coinciding with switches also
Example 2
Box 4 Code-switching and script-switching
This example of a Japanese make-up ad illustrates a type of script-switching
which is inherent in the Japanese writing system, and shows that certain
kinds of CS can be subsumed within what might otherwise be considered a
monolingual system. Japanese uses a combination of three scripts: Chinesederived characters called kanji, and two phonetic kana scripts called hiragana and katakana. Kanji are pictographic–ideographic characters used to
write nouns, adjectives and verbs. Hiragana are used to communicate the
grammatical functions of a statement, for example, verb conjugation/tense,
particles and adverbs. Katakana by contrast is used in contemporary
Japanese mainly to transcribe foreign words phonetically.
The three scripts may be switched so that words written in kanji, for
example, can be phonetically transcribed in either of the kana scripts.
Although convention and context require that a combination of scripts
should be used in standard texts, the three scripts offer great possibility for
experimenting with language, and depending on how the scripts individually and in combination are manipulated, one can play with meaning, the
relationship between sounds and characters, the relationship between words
and gender, class, age, ethnicity and so on.
As an example of how the scripts work in a standard setting, consider line
1 from this ad:
キュㅡト な 小箱
に、 パーリイ な グロス と 口紅。
na kobako ni,
na gurosu to kuchibeni.
* little box in,
* gloss
and lipstick.
Pearly gloss and lipstick in a cute little box.
* na – adjective particle
‘Cute’ and ‘pearly gloss’; are approximated phonetically into Japanese pronunciation using katakana (in italics). In contrast to the more formal kanji (in bold)
and functional hiragana (normal font), katakana in this context lends a lighthearted sense of fun, trendy youthful fashion and cute feminine style. (For more
information on the Japanese writing system, see Florence, 2003.)
Code-switching in conversation
includes quotation, addressee specification, interjection, reiteration, message
qualification and personalization v. objectification (see similar lists in SavilleTroike (1982), Valdès-Fallis (1977) and Zentella (1997). Markedness Theory and the Rational Choice Model Myers-Scotton’s
‘Markedness Model’ (1983; 1993a) represented an attempt to integrate
these functions into a more comprehensive model. She argued that in any
given social circumstances, a particular variety is the expected or “unmarked” –
i.e. the unremarkable – one. So, for example, switching to the local vernacular to
talk about home/family is “unmarked”, whereas switching to the local vernacular in a public speech is a “marked” choice. Myers-Scotton coins a new
“conversational maxim” (Grice, 1989) which corresponds with marked choices:
“Negotiate a change in the expected social distance holding between participants” (1993a:132).
Myers-Scotton set out to answer the question “What do bilingual speakers
gain by conducting a conversation in two languages (i.e. through CS) rather
than simply using one language throughout?” (1993b:3). She drew on numerous theories linking behavioural choices to social constraints, including that on
Power and Solidarity (Brown and Gilman, 1960), Politeness Theory (Brown
and Levinson, 1987), Speech Accommodation Theory (Giles and Smith, 1979;
Giles and Coupland, 1991) and Conversational Principles (Grice, 1989).
In a development of this theory, she presented bilingual speakers as “rational
actors” (Myers-Scotton, 1999; Myers-Scotton and Bolonyai, 2001), who are
constrained by social norms and conventions. In any given situation, they are in
a position to make linguistic choices, which may either conform with the
prevailing social norms and pass as “unmarked”, or, alternatively, redefine the
current set of “rights and obligations”. CS itself is the unmarked choice
in situations where two sets of identities are normally indexed simultaneously
in the community (this would not be possible under strict diglossia). CS can also
be an “exploratory” choice, when speakers are feeling their way to the most
advantageous way to conduct the conversation. This is the case in Example 3,
where a young man is trying to get a dance out of a young woman in a Nairobi
hotel. He starts off with the “neutral” choice, Swahili, but it is his switch to
English, following her lead, which seems decisive in persuading her to dance
with him.
Example 3
he: Nisaidie na dance, tafadhali.
Please give me a dance.
she: Nimechoka. Pengine nyimbo ifuatayo.
I’m tired. Maybe the following song.
he: Hii ndio nyimbo niyayopenda.
This is the song which I like.
she: Nimechoka!
I’m tired!
he: Tafadhali –
Please –
she: (interrupting)
Ah, stop bugging me.
he: I’m sorry. I didn’t mean to bug you, but I can’t help it if I like this song.
she: OK, then, in that case we can dance.
(Myers-Scotton, 1993b:146)
Li Wei (1998a) describes the application of Markedness Theory to CS as
arguably the most influential model since Gumperz made the distinction
between situational and metaphorical CS (see below). However, it places
the emphasis on the analyst’s interpretation of participants’ intentions rather
than on the – crucial – creation of meaning by participants within conversations. As we will see below, the notion that speakers make choices between
codes and code-switch in accordance with indexical values external to the
conversations and the speakers themselves has increasingly been regarded as
insufficient. Code-switching as “verbal action”: Auer and Li Wei In a critique of
the “Rational Choice” Model (2005), Li Wei demonstrates by means of
four extracts from conversations involving a young Chinese girl in England
that many of the turn-by-turn switches have a specific meaning which can be
interpreted within the conversation, not by reference to an external set of norms.
Gafaranga (2005) makes similar points in relation to a quite different linguistic
situation, that of Kinyarwanda-speaking Zairians in Belgium, for whom the mixing
of French and Kinyarwanda in itself already constitutes the medium of most
conversation. Therefore CS has to be identified on a case-by-case basis, when
speakers depart from the normal pattern of mixing in some fashion.
As Meeuwis and Blommaert (1994) have pointed out, in real conversational
interactions, it is not always possible to define the situation unambiguously in
terms of language choices, and speakers may pay attention to the participants’
conversational moves without referring to external precedents. Meeuwis and
Blommaert (1998) show that, within a community where CS is commonplace,
differences between “languages” may be much less salient than differences
between dialects, sociolects, speech styles, etc. In such circumstances, contrasting the two languages for a particular effect is just one way of using CS, and one
needs to look at the actual linguistic practices in any given community in order
to understand how the various (mixed) varieties are deployed.
A similar caution is expressed by Giles and Coupland (1991): “Speech is far
more likely to be dependent upon how speakers cognitively represent their
Code-switching in conversation
characteristics and subjectively define the scene than upon any objective classification imposed from without” (1991:15). Burt (2002) remarks that in a given
bilingual situation, the various maxims and sub-maxims within MyersScotton’s model can lead to opposite choices. For example, the “Deference”
Maxim dictates that one speaks the mother tongue of one’s guests to them, but
the “Virtuosity” Maxim dictates that one speaks the language which is understood by all the speakers present (p. 995). So if one is in the presence of guests,
along with others who do not share the latters’ mother tongue, the maxims are in
conflict and an ad hoc decision to prioritize one over the other has to be taken.
Sachdev and Bourhis (2001:415) point out that CS also occurs in the absence of,
or in spite of, a normative framework. At a methodological level, they observe
that the “maxims” are often defined by using examples of code-choices which
are then in turn re-cited as empirical evidence of those maxims, introducing a
certain circularity.
Auer (1998a; 2005), like Li Wei, considers CS to be part of “verbal action”,
and that its meaning can best be found at a level of conversational structure
which is neither grammatical nor dependent on larger societal structures,
although both of these are considered to have some relevance. In keeping
with the CA tradition, he looks for meaning in units going beyond the sentence,
and believes that the only way the analyst can prove that a given set of
co-occurring linguistic features is perceived by participants as a distinct code
is “by showing that switching between this set and another is employed in a
meaningful way in bilingual conversation” (1998a:13).
He takes an example of a conversation from Myers-Scotton. The farmer
speaks Lwidakho and some Swahili (italics), and is asking for some money.
Myers-Scotton interprets the divergent language choices as follows: the
worker shows his rejection of the farmer’s appeal for solidarity by using
varieties which are marked (“out-group”) in this context, English and
Example 4
1 farmer:
Khu inzi khuli menyi hanu inzalaAs I live here, I have hungerworker: [interrupting] Njaa gani?
What kind of hunger?
farmer: Yenya khunzirila hanuIt wants to kill me hereworker: [interrupting again, with more force) Njaa gani?
What kind of hunger?
farmer: Vana veruOur children-[said as an appeal to others as brothers]
worker: Nakuuliza, nja gani?
I ask you, what kind of hunger?
farmer: Inzala ya mapesa, kambuli.
Hunger for money.; I don’t have any
8 worker: You have got a land.
Una shamba.
You have land [farm]
Uli nu mulimi.
You have land [farm]
11 farmer: … mwana meru… my brother12 worker: … mbula tsisendi.
I don’t have money
Can’t you see how I am heavily loaded?
(Myers-Scotton, 1993a:82; Auer, 1998a:9–10)
Auer points out that the conversational “moves” made by the participants are
at least as significant as the conversation-external factors invoked by MyersScotton in explaining the language choices. While not denying the relevance of
the wider, “macroscopic” factors, he shows how an analysis of CS which
simultaneously pays attention to the sequential structures in conversation –
initial request, clarification request, elaboration, indirect decline, second
attempt at request, etc., can uncover patterns of convergence or divergence
between the speakers which would otherwise not have been apparent
(1998a:9–13). For example, in L8–10, the worker does in fact move from
English to Swahili to Lwidakho, thus achieving both emphatic repetition and
gradual convergence to the farmer’s variety. It is only in the third and last
request/decline sequence that the worker decisively diverges from the common
language and refutes the request.
Li Wei (1998a) makes a similar point and also contrasts Myers-Scotton’s
analysis with the CA approach, which focuses on speakers’ procedures for
arriving at local meanings. He shows how CS can be used to draw attention to
a new conversational move, regardless of the direction of the switch, and
analyses the productions of Chinese–English bilinguals in terms of Preference
Organization. In a rich database gathered in the Chinese community on
Tyneside in the UK, in which inter-generational talk is the prime site for CS,
he argues that it is not the structural contrast between Chinese and English
which provides an explanation for the switching patterns. A shift to English is
taking place over three generations, from Chinese monolingualism to Englishdominant bilingualism, and CS varies according to generation, network patterns
and the particular group studied (Li Wei, 1998b).
Example 5 shows how Li Wei’s subjects’ language choices simultaneously
reflect their linguistic preferences and authority structures in the family. Speaker
a addresses her mother in English, asking her to help with making a folder out
of cardboard. The failure of a to switch to her mother’s preferred language,
Cantonese, in spite of the latter’s repeated attempts to encourage her, are
Code-switching in conversation
indicative of a lack of cooperation which ends, as Li Wei points out, practically
in a communicative breakdown.
Example 5
[a is an eight-year-old girl, and c is a’s fifteen-year-old brother. b is their mother who is
in her forties.]
Cut it out for me (.) please
Cut it out for me (.) mum.
[Give us a look
Cut this out.
Give us a look.
Nay m ying wa lei?
You don’t answer me?
[To c] Get me a pen.
(Li Wei, 1998a:171–2)
The dual use of CS and pausing shows how CS complements or reinforces
discourse structuring devices which are available to monolinguals. This reinforcement function of CS parallels the use of such devices in combination with
one another in monolingual speech: for example the use of a discourse marker
like well in English is frequently accompanied by a pause – both before and
after – as well as a change in voice quality.
Another example of CS used to signify non-compliance occurs partly as
a result of the mismatch between the linguistic preferences of the younger
and the older generation: Li Wei found switching between turns in such
cases to be characteristic of “dispreferred” responses, as in the following
Example 6
[b, a twelve year-old boy, is playing on the computer in the living room. a is his mother]
a: Finished homework?
b: [2 second pause]
a: Steven, yiu mo wan?
want to review your lessons?
b: [1.5 sec. pause] I’ve finished.
The boy’s long pauses and his use of English are both indicative of his lack
of enthusiasm for telling his mother anything about his homework (Li Wei,
Code-switching compared with monolingual
conversational moves
In Gardner-Chloros, Charles and Cheshire (2000), a direct comparison is made
between the functions which CS has been shown to fulfil in bilingual conversations, and the equivalent expression of these functions in a monolingual
context. Two assumptions were made. The first is that bilingual code-switchers
and monolinguals accomplish basically the same conversational functions with
the different means at their disposal. Bell writes, for example: “having two
discrete languages available rather than a continuum of styles simply throws
into sharper focus the factors which operate on monolingual style-shifting. The
social processes are continuous across all kinds of language situations”
(1984:176). The second assumption is that, except in the case of “unmarked”
CS, it is more likely that switches are functional than non-functional. Whether
such switches are an instance of “rational choices” in Myers-Scotton’s terms, or
simply capitalize on contrasts within the conversation, we expected most
genuine switches be motivated.
In order to get round the problem of obtaining matched samples of monolingual and bilingual speech, code-switched and monolingual passages within
the same conversations were compared. This allowed a direct comparison to be
made of the way in which particular conversational effects are realized monolingually and through CS. The conversations analysed were between members
of a close Punjabi family/friendship network in London. Its members were
bilinguals, of varying degrees of dominance in English and Punjabi, and
habitual code-switchers. The use of discourse features in bilingual and monolingual conversations was compared, concentrating on four common types of
transition, which one would expect to be flagged both in monolingual and codeswitched conversational contexts. These were: (1) asides; (2) quotations; (3)
reiterations; and (4) “but” clauses. The occurrence of each of the four features
was noted and then classified as either monolingual or code-switched, according to whether the feature in question coincided with a change in language or
not. Examples 7 (a) and (b) illustrate the use of asides used monolingually and
bilingually in the same conversation.
In monolingual conversation, speakers often mark off “parenthetical”parts
of their utterance by a change in voice quality – in English often to a lower
pitch. This may be accompanied by other paralinguistic features such as a
gesture, facial expression or change in the direction of their gaze.
Example 7(a)
[Context: talking about one of the participants’ future son-in-law]
renu: What does he do? – to ask the usual question
Code-switching in conversation
The aside (underlined) is marked by lower pitch. The speaker is highlighting
a metaphorical change in addressee, as the aside is a metalinguistic comment
addressed to herself as much as to her interlocutor. The way in which such a
device may be used to mark the change from speaking to an interlocutor to
thinking aloud (and thus addressing oneself) can equally be seen in the following bilingual example:
Example 7(b)
[Context: talking about the speaker’s son’s friend)
1 umi: saada na, Charanpreet’s friend and er
our now Charanpreet’s friend and er
orna dtha munda siga
their of boy was
how old is Davan now?
eleven? So he’s about twelve
now our Charanpreet’s friend, they had a son, how old is Davan now? eleven?
So he’s about twelve
The speaker uses English and Punjabi in lines 1 and 2, and then switches to
English to pose the parenthetical question (underlined) in line 3. This is
marked by a change in voice just as in the previous, monolingual example.
Here the aside also serves a “fact-checking” function, in which confirmation
is implicitly sought from the interlocutor (Gardner-Chloros, Charles and
Cheshire, 2000:1317).
Dual marking of a discourse function through CS and voice tone was also
seen in relation to reiteration. Typically, repetition and reiteration achieve a
number of discourse functions simultaneously (Tannen, 1989; Coates, 1996). In
our corpus, the functions included emphasizing or clarifying certain stretches of
an utterance, and acting as a floor-holder. These functions have been identified
for reiteration in monolingual discourse. When reiteration coincides with CS,
however, the functions are marked twice over. Furthermore, switching
languages for repetition allowed speakers to hold the floor and to create
coherence between different parts of their utterance without the marked connotations of monolingual exact repetition, which can appear rude or
When CS occurred in quotation sequences, the additional dimension it can
bring to discourse was often very clear. It is a well-known function of CS to
frame a quotation – here the quote was sometimes framed twice over, once with
its quotative verb and then with the change in language. The use of CS not only
marked the boundary between the quotative verb and the quote itself. It also
gave speakers another “voice” in which they could encode expressive meanings. In monolingual discourse, such effects can be achieved by using a higher
proportion of a specific phonetic variant (Eckert, 1996; Eisikovits, 1991;
Holmes, 1995). A subtle change in the use of a single phonological or morphological variant, although in essence comparable to the use of CS as a “we-code”,
is presumably less salient to interlocutors than a change of language.
Finally, switches coinciding with but and its Punjabi equivalent par are
quintessential examples of the way that the speakers in this study marked textual
connections within and across their utterances twice over – once with the
contrastive conjunction itself and, simultaneously, with a contrast in language.1
When CS occurred with these conjunctions, additional effects of CS could be
observed: a precise mapping of form and content, implying a habitual action in
contrast to a single event, or as an aid to turn-taking. But can also have these
roles in monolingual discourse. The point is that the function is doubly marked
when CS occurs and, presumably, doubly salient to interlocutors as a result.
Example 8
[Context: Renu regretting that pictures of family in India were not evident in a home
video made by some of her British Asian relatives who recently visited India]
1 renu: me kya ki saria
nu fir dthek sugdthi
I said that everyone to then see can
kutho I haven’t been to India for a long time
3 umi:
oh yeah so they didn’t take the <family>
4 renu: <par> kisi
dthi family dthi ni eye vich
anyone of family of not come in
I said [to myself] that I’d be able to see everyone because I haven’t been to
India for a long time
6 umi:
oh yeah, so they didn’t take [ie. take pictures of] the family
7 renu: but nobody from the family was in it [ie. in the video]
Renu begins in Punjabi and then, in line 2, switches to English. Umi then
takes the floor in English. Then in line 3, Renu overlaps with Umi’s last word
(family) and switches back to Punjabi, signalling her reclaiming of the floor with
the Punjabi par ‘but’. At the same time, the use of ‘but’ in this position links her
utterance to her preceding turn, continuing the story about how, because she had
not seen the relatives in India herself for a while, she was disappointed that they
are not included in the home video. The language switch is therefore at the same
time both contrastive and cohesive (Gardner-Chloros, Charles and Cheshire,
2000:1333–1334). In a study of Mandinka – Wolof – English CS in the Gambia,
Haust similarly found that a high proportion (63.4 percent) of the insertions of
grammatical morphemes were formed by the two English words so and because
Oesch-Serra (1998) shows how the French and Italian words for ‘but’ have become complementary in terms of referential meaning in the bilingual speech of Italian migrants in Switzerland.
Therefore, which one is used at a particular juncture depends on the intended meaning and not on
the base language which is in use in that sentence or utterance.
Code-switching in conversation
Although the discourse functions achieved by CS can be performed monolingually, they are more salient when they are marked by CS, because they are
marked twice over. The discourse functions of CS have long been recognized,
but thanks to this direct comparison, we can see more clearly how these
functions relate to the features that occur in monolingual speech.
A Western view of “intentionality”?
A different view on ascribing motivations to CS is provided by Stroud (1992;
1998). Stroud studied CS in a non-Western context, between Tok Pisin (one of the
three national languages of Papua New Guinea) and Taiap, a language spoken in the
village of Gapun by a mere eighty-nine people. He pointed out the inapplicability,
in this context, of Gumperz’s “in-group/out-group” view of bilingualism, as he
claims that no domain, speech genre or topic is conducted exclusively in one
language. More fundamentally, he criticizes the whole notion of “intentionality”,
which he claims is based on an inappropriate, Western view of “personhood”.
Drawing on work by Duranti (1988), and Hill and Hill (1986), itself influenced by
the Bakhtinian notion of “voices” (or points of view on the world), he claims that
CS in Gapun should be seen as a series of rhetorical moves which highlight
contrasts, and/or perceptual shifts providing different points of view on a situation.
“The words of others carry with them their own expression, their own evaluative
tone, which we assimilate, re-work and re-accentuate” (Bakhtin, 1986:89). There is
little doubt that CS can provide different “voices” to the same speaker. This finding
ties in with research on reported speech in monolingual contexts: Tannen (1989),
for example, points out that most reported speech is “constructed dialogue”, i.e. has
never in fact been spoken.
Although Stroud is much closer to a CA interpretation than to one based on the
social connotations of the two languages like that of Gumperz or Myers-Scotton, he
is critical of CA as well, arguing that it is unclear what conversation analysts mean
by “language as social action”, and what the latter’s relationship is to other kinds of
(non-linguistic) action (1998:340–341). This is to some extent a general critique of
sociolinguistics for not making clear exactly how the structures described are
related to their social uses. He implies that only a deeply ethnographic approach
can get anywhere near understanding the “meaning” of CS from an emic perspective. In Gapun, meaning is not something the individual brings to the conversation
and tries to put across to others, it is a “negotiated product” which emerges from the
conversation. In such a case, the problem, he claims, of assigning meaning to codeswitches is first of all that of deciding: “Whose meaning is it?” (1992:151). Stroud’s
approach is illustrative of the diversity of approaches which CS has attracted, and
while his critique may be correct, he gives little indication as to how one should best
proceed in the light of it. It seems likely that all these levels of insight about CS have
their place and that the main challenge is to decide how they can be fitted together.
Accommodation, attitudes and audience design
Compared with other areas, there is a relative lack of work on the social
psychological aspects of CS, although the concepts used by social psychologists are extremely relevant to an understanding of it and such results which
exist should be seen as complementary to research on other aspects (Sachdev
and Bourhis, 2001).
Accommodation and audience design
CS is one of the possible ways of accommodating to the interlocutor’s linguistic
preferences (Coupland, 1985; Sachdev and Bourhis, 1990; Scotton, 1976; Wei
Zhang, 2005). It can serve as a compromise between two varieties, where these
carry different connotations or social meanings for speakers and interlocutors
(see above). It may also, of course, be the only possibility open to a speaker
where there is a mismatch between their level of competence in the relevant
languages and that of their interlocutor.
This compromise function – particularly where it allows the speaker to
address an audience made up of people with disparate linguistic competences –
is not limited to spontaneous speech. It is also exploited by politicians in their
speeches and comedians in their jokes (Woolard, 1988). Generally, it is used in
the media which are aimed at multilingual audiences for multiple functions.
For example, Hindi–English CS is extensively used in programmes on the
Asian cable television channel Zee TV. This channel aims to appeal to the
widest possible audience, including young second-generation Asians whose
main language is English, as well as to their parents’ and grandparents’ generation, whose main language may be one of a variety of Indian languages (e.g.
Punjabi, Gujerati), but who have Hindi as a language of literacy. This extract is
taken from a drama serial or soap shown on Zee2 (Gardner-Chloros and Charles,
2007). The scene opens with the male speaker pacing around impatiently, when
a well-dressed, older woman rushes in, obviously late for their appointment. By
reading the original and ignoring the translation, the non-Hindi-speaking reader
can experience the role of the CS as a prop to comprehension.
Example 9
Oh, hi
female: Ajay, I’m sorry, tyre flat horgya tha ….. [HINDI] ….
Ajay, I’m sorry, I got a flat tyre…. [Hindi] ….
Some monolingual Hindi stretches have been glossed over with the word HINDI, in order to
highlight the parts containing CS, where it can be seen how the few English words used provide a
useful prop, along with the visual action, to following the story.
Code-switching in conversation
Vese you can afford a driver, khud kyor drive kurthi he?
Mera mutlaab he it’s not safe
these days…. [HINDI] …..
Actually, you can afford a driver, why do you drive yourself?
I mean, it’s not safe these days…. [Hindi] …..
[HINDI] …. aap ko barri problem hothi howgi …. [HINDI] ….
..[Hindi] ….it must give you a problem….[Hindi] ….
Aap jesi progressive xxxx esi barth kurri he?
a progressive xxxx check? like you, saying things like that?
female: [HINDI] …
[several seconds’ pause] Oh [another long pause] I see
The last switch to English by the male speaker is punctuated by portentous
pauses, implying that the previous speaker’s utterance in Hindi was a significant
revelation which explained her late arrival in a more thorough-going fashion
than the flat tyre could do. Thus the CS here is both a compromise to suit an
audience with various levels of competence in English and Hindi – many
themselves code-switchers – and simultaneously functional within the conversation itself.
Local radio stations aimed at immigrant communities are also often rich
seams for CS. This is the case for London Greek Radio, aimed at the London
Cypriots, and also of Beur FM, a French community station aimed at the North
African population. Aitsiselmi (2003) has identified pretty well every known
category of CS on this station, including inter-sentential, intra-sentential,
flagged, identity-related, unmarked and humorous or ludic. As an example of
“unmarked” CS with no apparent function, he reports the following:
Example 10
Ils sont revenues laHqu
juste m3a
la grève des huit jours
They came back they arrived just with
the one week strike
tal3u fi
Air France a sidi, ki diKularte l’avion avait des problèmes. Umba3ed,
so they took Air France When the plane took off the plane had problems.
l’équipage kifash y-anunsiw la nouvelle
les passagers?
How was the crew to announce the news
the passengers?
(Aitsiselmi, 2003:4)
Another major function of CS is what Gumperz called “addressee specification”.
Using the appropriate language to address different interlocutors allows the
participants to continue the conversation smoothly, without undue flagging of
who they mean to address (“flagged” switches involve inserting a conversational
marker or comment at the point where the switch occurs). This is common in
Alsace, where balanced bilingual speakers of the middle generation often use CS
in this way. For example in the Beck family, the following conversation took
place over a meal between the parents, Mr and Mrs Beck. They are balanced
bilinguals but generally speak to one another in Alsatian. The grandfather is
markedly more at ease in Alsatian, and the Becks’ two teenage daughters are
basically French speakers. CS allows Mrs Beck to minimize the confusion which
could be caused by her holding two conversations at once, one with her aged
father about the weather, and the other with her husband and daughters
about serving the meal.
Example 11
mr beck:
Awwer hit morje isch kalt gsinn, oh kalt!
But this morning it was cold, oh so cold!
mrs beck:
’s isch kalt, ’s isch Winter!
It’s cold, it’s winter!
mr beck:
A table, les enfants!
Lunch is ready, children!
mrs beck:
De Putzlumpe isch gefrore sinn. Allez vite maintenant, allons!
The mop was frozen. Come on now, quickly!
grandfather: Hit morje haw ich gelööjt, sinn’s vier Grad g’sinn.
I had a look this morning, it was four degrees.
mrs beck:
Ja, s’isch unter nul gsinn. S’isch g’frore g’sinn. Des pommes de
terre, qui veut des pommes de terre?
Yes, it was below zero. Potatoes, who wants potatoes?
(Gardner-Chloros, 1991:114)
Addressee specification of this type overlaps with accommodation. As we
have seen, however, the concept of “audience design” as such has mainly been
applied to monolingual style-shifting. Specific attempts to investigate accommodation in the context of CS have been made by Lawson-Sako and Sachdev
(1996; 2000). The definition of CS used in these studies should be noted, in that
the object of their study is in fact language choices. In a series of anonymous
roadside surveys in Tunisia, researchers of different ethnicities posed as
passers-by asking the way to the station. They used Tunisian Arabic, French
or a combination of the two. Results indicated that subjects converged most
readily to the linguistic stereotype of in-group usage when the questioner looked
as if they belonged to the in-group (i.e. when they were Arab as opposed to
European), whether or not the questioner addressed them in the in-group
language. The “unmarked in-group code which most accurately represents the
bilingualism of the country” turned out to be CS between Tunisian Arabic and
French (2000:1358).
If the desire to accommodate sometimes manifests itself through CS, as the
above study seems to show, then we should bear in mind that accommodation is
itself a more complex phenomenon than might appear at first sight, and that like
CS, it is not found equally across all members of a population. Woolard (1997)
found, in a survey of Catalan adolescents, that girls reported a greater tendency
towards linguistic accommodation to their interlocutor than boys did. Catalandominant female speakers maintain Catalan with Castilian speakers, but converge by using Castilian if they consider the Castilian speakers as their friends.
As Woolard observes, for girls, “social distance allows one to carry on a
Code-switching in conversation
bilingual conversation, while closeness and friendship demand a shared language … For the girl, solidarity demanded accommodation and social distance
allowed linguistic mismatch. For the boy, social distance demanded accommodation, and solidarity entitled individuals to use their own languages” (p. 549).
A monolingual parallel to this can be found in Brown (1994), who found that
women show more positive politeness (PP) than men towards friends, and more
negative politeness (NP) in public.
Attitude studies of CS are still relatively few and far between, and most of our
information on this is gleaned from a variety of studies where responses are
elicited about attitudes along with other aspects. An early exception to this was
Chana and Romaine (1984), who reported negative attitudes towards CS among
Punjabi–English bilinguals in Birmingham, in spite of their almost exclusively
using a CS mode. Bentahila (1983) carried out a matched guise experiment
among 109 Arabic–French bilinguals in Morocco, and found a large majority
expressing disapproval of the CS guise, with attitudes ranging from pity to
disgust. Gibbons (1987) found similar reactions among Cantonese–English
code-switching students in Hong Kong, although he also identified an element
of covert prestige associated with it. Gumperz’s subjects attributed CS to
characteristics such as lack of education, bad manners or language inability
(1991), and Zentella’s to language deficiency, rather than language skill or
discourse needs (1997).
Lawson-Sako and Sachdev (2000) report on a matched guise among 169
Tunisians in which they found CS to be rated lowest of all the guises which were
used, regardless of the speaker’s gender, although gender did affect reactions to
the other varieties. Nevertheless, in a follow-up language diary study, the use of
CS was widely reported in a range of circumstances (2000:1353–1354).
McCormick (2002) found a variety of attitudes towards CS within the same
South African community.
A study of London Greek Cypriots revealed reasonably positive attitudes
towards CS in this community, with some variation depending on education,
occupational group and age (Gardner-Chloros, McEntee-Atalianis and Finnis,
2005). Subjects from lower occupational groups had the most favourable
attitudes towards CS; in fact, the more educated the respondents, the less
favourable their attitude towards CS. The younger respondents disapproved
less of CS, and saw it as more advantageous, than the older ones. Several
significant differences were found between attitudes – and in reported CS
practices – among Cypriots in London and in Cyprus itself, although the
language combination is the same and several significant cultural values are
shared. This was ascribed to the languages having a different “market value” in
Cyprus and in London (Bourdieu, 1997). Overall, it seems that CS is gradually
gaining acceptability in many contexts, as cultural, racial, musical, culinary and
other types of hybridity or “fusion” are better tolerated and indeed become
Much work remains to be done, however, for a full understanding of attitudes
to CS. Results such as those of Aikhenvald (see Chapter 2) and Bentahila (see
above), where extremely negative attitudes are expressed – often at odds with
people’s behaviour – suggest that attitudes to CS are learned rather than
spontaneous. In order to get a picture of people’s spontaneous reactions, one
should look at contexts where speakers are in no way insecure about their
language use, where they do not feel their language to be threatened and, at
the same time, where they are not taught or indoctrinated to believe in “purity”
as a linguistic ideal. Unfortunately, the world can be roughly divided into the less
developed areas, where the first condition applies, and the more developed world
where the second one does (Le Page and Tabouret-Keller, 1985:181–182).
Gender is considered one of the most important sociolinguistic categories.
Studies of the interaction of gender with linguistic performance have become
increasingly subtle, avoiding the facile generalizations of the 1970s studies.
Gender has assumed more prominence within the discipline rather than less, as
the ways in which it is studied have become more diversified.
The research surveyed below shows that CS cannot be correlated in any
direct way with gender, but intersects with a large number of intervening
variables which are themselves connected with gender issues. Following this,
a piece of research is presented in greater detail (Gardner-Chloros and Finnis,
2004) which shows how CS is woven in with female discourse strategies and
discourse needs, via the notion of politeness (Brown and Levinson, 1987).
Code-switching and gender in various communities
The long-established finding that women use more standard forms than men
(Labov, 1972; Trudgill, 1972; Chambers, 2003) derives from monolingual
settings. In its simplest form, it can usefully be tested in bilingual contexts.
First, we need to know whether, in a given case, the choice of one or the other
variety corresponds with a choice between the vernacular and the prestige code.
In some cases, it is the CS mode itself, as we have seen, which carries the “ingroup” connotations and may be considered the “local” type of speech (Swigart,
Given the generally negative judgements of CS outlined above, a study was
carried out to find out whether the widespread finding that women use more
Code-switching in conversation
standard, and less non-standard, language than men was reflected by a clear
gender difference in the amount of CS they used. The finding would gain
support if women were found to code-switch substantially less than men
(Cheshire and Gardner-Chloros, 1998). Transcribed recordings from two immigrant communities in the UK, the Greek Cypriots and the Punjabis, was used to
test the hypothesis. The results were negative – there were no significant
differences between men and women in either community regarding the use
of any kind of CS, though there were substantial differences between the two
communities, both as regards quantity and type of CS.
Some other studies, on the other hand, have found differences in either the
amount or the type of CS used by women and men within the same community (Poplack, 1980; Treffers-Daller, 1992). In a study in the Gambia, Haust
(1995) found that men used CS twice as much as women, especially using
discourse marker insertions, whereas women tended to change varieties outside the turn unit.
Such differing findings in different communities should come as no surprise,
given the shift which has taken place within language and gender studies from
essentialist to constructionist views (Winter and Pauwels, 2000). As Swigart
(1991) argued, women, even within a given society, do not all behave as a
monolithic group. Gender is not a fixed, stable and universal category whose
meaning is shared within or across cultures. It cannot be separated from other
aspects of social identity and its meaning varies in different domains: “A nonessentialist view sees gender as a dynamic construct, which is historically,
culturally, situationally and interactionally constituted and negotiated”
(Winter and Pauwels, 2000:509). Conversely though, the variety within these
findings should lead us to relativize the usual pattern of sex differentiation,
which Chambers (2003) referred to as a “sociolinguistic verity”. This can come
about if we look not only at statistical information about how many instances of
variant X are produced by women or men, but at the discourse context and the
reasons why particular choices are made (see Stroud above).
Furthermore, use of particular linguistic forms does not always signal the
same underlying motivations. Traditionally polite or indirect forms do not
necessarily indicate underlying compliance. Brown (1994) found that in
Tenejapan society, even when women are not being polite in essence, characteristic female strategies of indirectness and politeness are nevertheless manifested
in their speech. Brown suggests that this might help us make sense of the finding
that women appear more cooperative than men in interaction. While cooperative strategies are being used, what is being achieved may be opposition and
disagreement. The way in which this is done in particular instances, the
strategies which are typical of women or of men in specific communities, and
the particular types of discourse where CS is brought to bear, are often associated with different sexes in a given community.
Code-switching, gender and politeness
In Gardner-Chloros and Finnis (2004), the link between language and gender
was explored by considering whether certain specific functions of CS are more
common among women or men in the Greek Cypriot community. Various
findings were taken up from Cheshire and Gardner-Chloros (1998), mentioned
above. The earlier study did not eliminate the possibility that, although the
overall switching rate between the sexes did not differ significantly, women and
men were code-switching for very different purposes.
Two sets of data were used, thirty interviews carried out in the London Greek
Cypriot community (Gardner-Chloros, 1992) and transcriptions of recordings
carried out at meetings of a Greek Cypriot youth organization. These meetings
were informal, and took place at a range of venues, including a community
centre, a coffee shop and the home of one of the participants. The participants
were five males and five females between the ages of twenty-three and twentynine who had all completed higher education.
Sifianou’s (1992) comparative study of politeness in England and Greece proved
particularly useful. It was pointed out there that different cultures place emphasis on
different values, which values are moreover interpreted differently. Basing her work
on Brown and Levinson’s (1987) theory of Positive and Negative Politeness,
Sifianou argued that “Politeness is conceptualised differently and thus, manifested
differently in the two societies; more specifically that Greeks tend to use more
positive politeness devices than the English, who prefer more negative politeness
devices” (1992:2). It is not the case that some cultures or societies are more polite
than others. The difference is the quality, rather than the quantity of politeness
strategies, in that speakers are polite in different, culturally specific, ways.
For example, Greek speakers are more direct when it comes to making
requests, when giving advice or making suggestions. The cultural norm in
England requires a more distant code of behaviour, and requests, among other
speech acts, are expressed more elaborately and indirectly. Sifianou argues that,
in England, requests are perceived to a greater extent as impositions, and as such
need to be accompanied by more elaborate politeness strategies. Therefore a
variety of options are available to the interlocutor when making a request,
allowing the imposition created by the request to be minimized, e.g. You don’t
have a pen, do you? (p. 140).
In contrast, Greeks define politeness in very broad terms. Sifianou found that
their definition included attributes which might be better described in English in
terms of “altruism, generosity, morality, and self-abnegation” (p. 88). Greeks
reported that “a warm look, a friendly smile, and in general a good-humoured
disposition and pleasant facial expression are integral parts of polite behaviour”
(p. 91). Her overall message is that English culture values distance, and Greek
culture values intimacy.
Code-switching in conversation
This is supported by several examples in Gardner-Chloros and Finnis (2004),
which indicate that, when being direct, Greek Cypriot speakers prefer to switch
to Greek, as directness is more acceptable in Greek culture. This seems especially to be the case for women, of whom, as in many Western societies, there
is an expectation that they will be more polite and consequently more indirect
than men. At the same time, because Greek is a more positively polite language,
when being intimate, speakers may also prefer to use Greek. Similarly, Zentella
(1997) notes that in the Puerto Rican community in New York, commands are
often repeated in Spanish, after being delivered in English, in order to soften
their impact/harshness.
Three of the functions which are noticeably associated with CS, which were
labelled humour, bonding and dampening directness, are illustrated below.
There are significant overlaps between the three, which reinforces the idea
that there is a general politeness function associated with CS. For different
reasons, which are discussed in each case, it was considered that these uses of
CS were particularly typical of women in the community, though by no means
exclusive to them. Code-switching used for humour
Example 12
1 m1:3 …??? happen to know anyone that has like a colour laser jet..
2 f1: I know a place where they do???
3 m1: yeah
4 f1: ???
5 m1: what make are they?
6 f1: En iksero, en leptomeries
I don’t know, these are details
[general laughter].
(Gardner-Chloros and Finnis, 2004:524)
Speaker F1 is relying on her interlocutors’ familiarity with Greek culture, in that
she adopts the “voice” of a particular Greek stereotype, that of a laid-back type
who won’t bother with too much detail (line 6). The fact that she is “playing a
part” is indicated by a change in voice quality for the remark in Greek. In this
way, she justifies her ignorance of the technical details of the photocopier by
bringing in a persona who represents this particular Greek attitude. Code-switching used for bonding CS was often used to indicate identification or intimacy. In the following example, the speakers are talking about a
conference they are organizing. Speaker F1 suggests the topic of arranged
In this study M1 indicates the first male speaker, M2 the second male speaker, F1 the first female
speaker, etc. ‘???’ indicates inaudible speech.
marriages, a traditional aspect of Greek Cypriot culture. She refers to her own
mother’s concern about her finding a husband and getting married:
Example 13
1 f1: Am I the only person that gets??? by their parents already?
2 m1: What, about getting married?
3 f1: Yeah, she started today.
4 f2: ???mana sou?
your mother?
(Gardner-Chloros and Finnis, 2004:525)
In line 4, Speaker F2’s intervention in Greek can be viewed as an act of
positive politeness, or identification with F1, as another female Greek Cypriot.
She uses the language of the culture in which such traditional maternal attitudes
towards the marital status of daughters prevail. Gender therefore plays an
important role in this switch. Whilst the topic of marriage within the community
is relevant to all its members, it has much greater consequences for women, and,
as such, requires more positive politeness strategies in order to indicate
solidarity. Code-switching used for dampening directness In example 14, speaker
F1, after asking the same question in English twice and failing to get a response
from speaker M1, switches to Greek to elicit a response. Having succeeded in
doing so, she then switches back to English.
Example 14
m1: All right
f1: Stop, how many days is the conference?
m1: Guys, I wanna finish at seven o’clock
f1: I’m asking ! How many days is the conference?
m1: ??? It’s half past six.
f1: Kirie Meniko, poses imeres ine?
Mr Meniko, how many days is it?
m1: It will be around four days, I imagine
f1: Ok, four days, good … and what time?
(Gardner-Chloros and Finnis, 2004: 527)
The potentially face-threatening act – an escalation of repeated questions which
had been phrased pretty directly from the beginning – is carried off thanks to the
switch to Greek, which not only allows greater directness but is also the we-code
and the language of humour. CS is shown to offer a powerful toolkit for women
in the community, who can get away with jokes, strong repartee, etc. without
appearing aggressive or unfeminine.
Among the London Greek Cypriots, women seemed to make use of these
strategies to get round some of the traditional constraints on female discourse,
such as the expectation that it will be less forceful, pressing or direct than that of
Code-switching in conversation
men, or that making jokes is unfeminine. Women also use CS for solidarity in
certain contexts which are directly relevant to them, e.g. in talking about
mothers and their attitudes towards their daughter’s marital status. It would
not be surprising if, being more directly concerned, women talked about these
issues more than men, and so had occasion to use these PP strategies to a greater
extent, though this is obviously an empirical question.
To the extent that one can show that gender differences are contingent upon
culturally determined norms, the role of gender as such is relativized. It is shown
to be mediated by other factors, such as the power relationship between the
speakers and the conventions governing behaviour – which of course include
gendered behaviour – in the community. “We must criticize explanations of
difference that treat gender as something obvious, static and monolithic, ignoring the forces that shape it and the varied forms they take in different times and
places … Feminism begins when we approach sex differences as constructs,
show how they are constructed and in whose interests” (Cameron, 1992:40).
In Chapter 3 and in this chapter, we have considered CS from a sociolinguistic
perspective. In Chapter 3, we saw how CS could reflect social realities, from the
dominance relations between language groups to differences within groups, for
example between generations. The importance of making comparisons between
different CS situations was stressed, and illustrated by instances where such
comparisons were made. In this chapter, we have seen how the uses of CS are
woven in a complex fashion with various social motivations, from the desire to
accommodate to one’s interlocutor to the expression of gender identification.
There is rarely a simple, one-to-one correspondence between such factors and
the use of CS, even if – naively – one treats CS itself as a simple, on-or-off,
phenomenon. As Nilep has put it, CS is “the practice of individuals in particular
discourse settings. Therefore, it cannot specify broad functions of language
alternation, nor define the exact nature of any code prior to interaction. Codes
emerge from interaction, and become relevant when parties to discourse treat
them as such” (2006:63). Cashman (2005) also describes how social structures,
social identities and linguistic identities are all “talked into being”, and are
alternately constructed, accepted and rejected within the same conversation.
The participants’ role-taking is much more subtle than their simple social
identities (in this case, as Chicano/Latinos and Anglos) would suggest.
Authors including Auer, Li Wei and Gafaranga have all shown how the indexing of varieties with particular values is subordinate, or at least complementary,
to the particular uses which bilinguals extract from CS in their conversations.
The ways in which those varieties are manipulated by bilinguals are so
varied, and the linguistic configurations are so diverse, that it is tempting to
think that there are no limits at all to how languages can be combined, or to what
end. In some senses this is true. As we saw, borrowing, for example, arises at all
linguistic levels, not only the lexical level with which it is often associated.
Nevertheless, synchronically speaking, individuals are limited by their knowledge and perceptions and the ways in which they find it possible – or
acceptable – to combine them. The pragmatic approach to CS is marked by
the struggle between attempts to demystify and systematize – such as MyersScotton’s model – and the realization that, having once constructed such
systems, speakers can then turn round and deliberately ignore them or subvert
them, in their online productions, for their communicative ends.
In the next two chapters, we will look at two different aspects of the combining which takes place in CS, first from the point of view of grammar, and then
from the point of view of how linguistic elements, which may or may not belong
to separate “systems”, interact in the brain. Both these factors can in theory set
the outer limits on the possibilities of code-switching. But both grammatical and
psycholinguistic limitations rest first and foremost on our understanding of
what constitutes a separate variety or language in the first place, a question
raised in Chapter 1, and which impacts on every aspect of the study of CS.
Box 5 Code-switching in antiquity
CS was common all over the ancient world where Latin and Greek were in
contact with local language. This has been discussed in two recent volumes
(Swain, 2002; Adams, 2003).
The Roman orator Cicero, who styled himself as a philhellene, wrote a
series of letters in the first century ad to his friend Atticus (Cicero, 1999),
code-switching copiously from Latin to Greek (Adams et al., 2002). The
notable feature of these letters is that they are gossipy in style and were not
intended for publication. It is therefore reasonable to assume that the friends
code-switched into Greek when speaking, as Cicero does regularly in the
letters, as shown in three extracts below (with Shackleton Bailey’s
325 (XIII.18)
Scr. In Arpinati IV Kal. Quint. An. 45
<cicero attico sal>
Vides propinquitas quid habet, nos vero conf<ic>iamus hortos. colloqui
videbamur in Tusculano cum essem; tanta erat crebitas litterarum. sed id
quidem iam erit. ego in-terea admonitu tuo perfeci sane argutolos libros ad
Varonem, sed tamen exspecto quid ad ea quae scripsi ad te primum qui
intellexeris eum desiderare a me, cum ipsehomo πολυγρώτατος numquam
Code-switching in conversation
me lacessisset; deinde quem ζηλοτυπει n½ <intellexeris. quod si non
Brutum. multo Hortensium minus aut eos qui de re publica loquuntur.
plane hoc mihi explices velim in primis, maneasne in sententia ut mittam
ad eum quae scrips an nihil necesseputes. sed haec coram.
325 (XIII.18)
Aprinum, 28 June 45
<cicero to atticus>
You see the virtue of propinquity. Well, let us secure a property in the suburbs.
When I was at Tusculum it was as though we talked to one another, letters
passed to and fro so rapidly. But it will soon be so again. Meanwhile I have
taken your hint and finished off some neat little volumes addressed to Varro.
None the less I am awaiting your answer to my questions: (a) how you
gathered that he coveted a dedication from me, when he himself, extremely
prolific author as he is, has never taken the initiative, and (b) whom you
gathered him to be jealous of: if it’s not Brutus, much less can it be
Hortensius or the speakers on the Republic. The point above all which I
should really be glad if you would make clear to me is whether you hold to
your opinion that I should address my work to him or whether you see no
need. But we shall discuss this together.
391 (XV.16)
Scr. Asturae (?) III Id. Iun. an. 44 (?)
<cicero attico sal>.
Tandem a Cicerone tabellarius; sed mehercule litterae πεπινωμένως
scriptae, id quod ipsum προκοπὴν aliquam significaret, itemque ceteri
praeclara scribunt. Leonides tamen retinet suumillud ‘adhuc’; summis
vero laudibus Herodes. quid quaeries? vel verba mihi dari facile patior in
hoc meque libenter praebeo creduum. tu velim, si quid tibi est a tuis
scriptum quod pertineat ad me, certiorem me facias.
391 (XV.16)
Astura (?) 11 June 44 (?)
<cicero to atticus>
At last a courier from Marcus! But upon my word the letter is well written,
which in itself would argue some progress, and others too send excellent
reports. Leonides however sticks to his ‘so far’; but Herodes is enthusiastic. Truth to tell I am not unwilling to be deceived in this case and gladly
swallow all I’m told. If there is anything in letters from your people which
concerns me, please let me know.
392 (XV.16a)
Scr. Asturae (?) prid. Id. Iun. an. 44 (?)
<cicero attico sal>.
Narro tibi, haec loca venusta sunt, abdita certe et, si quid scribere veils, ab
arbitris libera. sed nescio quo modo ο'^iκος f ίλος. itaque me referent pedes
in Tusculanum. Et tamen haec ·ωπογραfία repulae videtur habitura
celerem satietatem. Equidem etiampluvias metuo, si Prognostica nostra
vera snunt; ranae enim ·ητορεύουσιν tu, quaeso, fac sciam ubi Brutum
nostrum et quo die videre possim.
391 (XV.16)
Astura (?) 12 June 44 (?)
<cicero to atticus>
The district, let me tell you, is charming; at any rate it’s secluded and free
from observers if one wants to do some writing. And yet, somehow or other,
‘home’s best’; so my feet are carrying me back to Tusculum. After all I
think one would soon get tired of the picture scenery of this scrap of
wooded coast. What is more, I am afraid of rain, if my Prognostics are to
be trusted, for the frogs are speechifying. Would you please let me know
where and what day I can see Brutus?
Grammatical aspects of code-switching
Grammatical approaches have been one of the most prolific sub-fields in the
study of CS, and there a number of useful summaries of work in this area (e.g.
Bhatia and Ritchie, 1996; Muysken, 2000). The volume of work on CS from
this perspective derives from the light which it can cast on grammatical theories:
“When sentences are built up with items drawn from two lexicons, we can see
to what extent the sentence patterns derive from the interaction between these
two lexicons” (Muysken, 1995:178). The purpose of this chapter will not be to
give a full picture of the theories, since these have been described thoroughly
elsewhere (Myers-Scotton, 1997; Muysken, 2000). Instead there will be a brief
overview, the emphasis being put on the role which grammar can play in our
understanding of CS – i.e. a reversal of the normal formula. It will be argued
that the grammatical work done on CS so far has failed to take the variations in
code-switching behaviour sufficiently seriously. Rather than being peripheral
issues, which can be treated under the heading of exceptions or instances of
some other phenomenon such as borrowing, the variations in CS grammars
should be taken as the central issue to investigate. Grammatical explanations
of CS will be more satisfactory to the extent that they can take on board the
variations related to differences in competence between speakers, differences
related to typological factors and differences due to sociolinguistic parameters –
all of which have often so far been regarded essentially as an inconvenience.
This is a product of the fact that we have so far lacked a general debate as to
how and why the notion of grammar can apply to CS in the first place (GardnerChloros and Edwards, 20041). The different senses of “grammar”, like those
of “language” (Le Page, 1989), do not all lend themselves to characterizing
performance data. There are at least three potential problem areas:
(1) A grammar is essentially a linguist’s description of properly formed
sentences, and hence represents an abstraction over a set of data. In most
types of grammar, the sentence (or clause) represents the upper limit. But
The discussion here derives largely from that paper.
the sentence – and its component parts of speech, nouns, verbs, etc. – are not
necessarily appropriate units for the analysis of spontaneous speech. Second,
grammatical approaches which only seek to explain CS within the sentence
can only account at best for some of the data, as much CS occurs between
sentences and between conversational moves (Auer, 1998a:3).
(2) There are doubts about a key concept used systematically in much of the
literature on CS grammars, namely that of the “Base” or “Matrix” Language
(actually a base grammar). The assumptions underlying this notion are
discussed below, (and more fully in Gardner-Chloros and Edwards, 2004).
It is argued that a misplaced faith in the role of the Matrix Language
underlies the failure of many grammatical proposals to account fully for
CS data.
(3) Much grammatical work on CS is based on an assumption that bilinguals
alternate in some meaningful way between two clearly distinguishable sets
of rules. But this question is not one which can be decided by grammatical
analysis alone, or the procedure is circular. As Alvarez-Cáccamo wrote
In order to argue convincingly for or against the existence of “code-switching
constraints” and “code-switching grammars” based on the two monolingual ones
(…), research should first convincingly prove that (a) speakers who code-switch
possess two (or more) identifiable linguistic systems or languages, each with its
identifiable grammatical rules and lexicon; and (b) “code-switched” speech results
from the predictable interaction between lexical elements and grammatical rules
from these languages. None of these assumptions, I believe, is proven yet. (AlvarezCáccamo, 1998:36)
Before looking at how grammatical theories have actually been applied
to CS, we will look briefly at some of the different senses in which the term
“grammar” is used in this research.
Types of “grammar”
Prescriptive/Pedagogical grammar
Linguists working on the grammar of CS are certainly not consciously putting
forward a prescriptive model of bilingual speech; they are attempting to uncover
universal regularities which underlie it. Nevertheless, a prescriptive element
can creep in: the outcome of specifying a “grammar” of CS is that there appears
to be a right and a wrong way to code-switch, or at least a “possible” and an
“impossible” way: it is said that certain combinations of words from different
varieties are not CS (see the notion of nonce loans discussed below in 5.3.1,
but also Myers-Scotton’s notion of “classic” CS). Some types of CS are seen as
“more equal than others”, and many grammatical approaches avoid engaging
with the variations between what is prevalent in different sub-groups, e.g.
Grammatical aspects of code-switching
different generations, or the same speakers in different contexts. There is a
parallel here with dialect/creole-speaking settings, where speakers often say of
their speech-mode that “it has no grammar” – a recognition that in non-focused
contexts, speech is not regulated by the rules which apply to standard, especially
written, varieties, and may therefore vary considerably both inter- and intraindividually.
Chomskyan/Universalist grammar
Chomskyan Universal Grammar (UG) is not actually a grammar as such, but
rather a “metagrammar”: it determines the forms that rules of individual grammars can take. It is therefore at several removes from the form of CS utterances.
The nature of the principles and constraints formulated within UG theory are
generally highly abstract, and liable to change in the light of developments
within the theory, the current version being Minimalism (MacSwan, 1999;
2005; Bhatia and Ritchie, 1996).
Chomsky (1986) made a distinction between E-language, meaning the totality
of utterances that can be made in a speech community, and I-language, defined
as “some element of the mind of the person who knows the language”. He
considered CS to be impure, as it does not represent a single set of choices among
the options allowed by Universal Grammar. Muysken (2000) offered various
possible explanations as to how there may not be a one-to-one correspondence
between the E- and the I-language: bilinguals combine modules from different
languages, and several E-languages may correspond to a relatively coherent
I-language. George (1990) further distinguished between what speakers know/
believe about their grammar, and how these beliefs are actually internally represented (“psychogrammar”).
In fact, the majority of grammatical studies of CS, including those of Poplack
and Myers-Scotton, attempt to explain surface-level regularities. Nortier
(1990:169–170) notes this important contradiction in their formulation: on the
one hand it is stated that in CS, syntactic rules of either “language” must not
be violated, which implies that underlying structures are the focus of attention;
on the other hand, the examples given are all to do with points at which the
surface structures are the same or different in the two languages. She shows
how, in Spanish–English, this contradiction does not raise serious problems,
whereas in her Dutch–Moroccan Arabic data, it does. Similarly, Clyne (1987)
and Romaine (1986) had suggested that the Government model failed to explain
the sites of CS because these are surface-structure properties. As Chomsky
wrote, “there is no reason to expect uniformity of surface structures … Insofar
as attention is restricted to surface structures, the most that can be expected is
the discovery of statistical tendencies, such as those presented by Greenberg
(1963)” (1965:118).
Formal grammars
These are generative grammars, typically expressed in a rigorous phrase-structure
formalism, which provide highly explicit grammatical descriptions of particular
languages. Cowie (1999) argues that even within generative discourse, the term
“grammar” is used to refer to (at least) three distinct objects:
the system of rules determining well-formed strings in a given language;
“competence” grammar, suppposedly consisting of a set of rules and principles
claimed to exist in the speaker’s mind; and
linguist’s model of the grammar.
The linguist’s model of grammar and the content of the mental grammar are
only indirectly related. Moreover, CS poses additional problems, in that it
often displays grammatical forms which are “hybrids” of the two monolingual
grammars (for example, double morphology). Such phenomena, also attested
in second language acquisition, do not necessarily undermine the search for
UG-type principles of syntactic organisation, but they do suggest that bilinguals
may be operating processes which cannot be explained by generative models.
Cognitive/Functional/Word grammars
The common feature of this cluster of frameworks is that they do not recognize
strict divisions between grammar/syntax, meaning and discourse functions.
For example, Myers-Scotton’s “Production approach” (1993b) attempts to tie
in the psycholinguistic notion of “activation” with the grammatical form of CS
productions, and in order to do this relies on the notion of “Matrix Language”
(see below).
Grammatical studies of CS have on the whole been based on the second and
third types of grammars. Studies based on Sense 2 (e.g. Di Sciullo, Muysken
and Singh, 1986), which seek to demonstrate universal patterns in CS, have so
far not succeeded in doing so. Many other studies have been based, implicitly or
explicitly, on grammar in the spirit if not the letter of the third type – statements
about the structure of particular languages, and how the differences between
them are reconciled in CS. In this case, the productions of bilingual speakers
are interpreted through the template of a set of regularities derived from a quite
different set of data, which is monolingual – and often introspective – and has
provided what is considered to be “the grammar” of Language X and that of
Language Y.
The application of grammatical models to code-switching
Broadly speaking, one can distinguish three major trends in the grammatical
study of CS:
Grammatical aspects of code-switching
(1) Variationist approaches: In the 1970s and 1980s, various attempts were
made to formulate grammars based on universal constraints on where CS
could occur in the sentence (Timm, 1975; Pfaff, 1979; Poplack, 1980; Sankoff
and Poplack, 1981).
(2) Generativist approaches, which took off in the 1980s (Di Sciullo, Muysken
and Singh, 1986; Joshi, 1985, Mahootian, 1993; Belazi, Rubin and Toribio,
1994; MacSwan, 1999; 2000; 2005).
(3) Production approaches (de Bot and Schreuder, 1993; Azuma, 1996; MyersScotton, 1993b). Since these concern psycholinguistic aspects of CS, they
are mainly dealt with in Chapter 6. Myers-Scotton’s Matrix Language Frame,
however, is presented as a direct alternative to variationist and generative
approches and is therefore considered below. It has been the subject
of repeated amendment by Myers-Scotton and her colleagues (see MyersScotton and Jake, 2009).
The variationist approach
When CS was first studied systematically at a grammatical level, it was quickly
observed that switches did not occur at random points in the sentence. Certain
kinds of switching, for example lexical switching, were very common, at least
in the Spanish-speaking communities in the USA which were originally studied,
whereas others, such as switching between a pronoun subject and a verb, were
very rare. This led researchers, for example Timm (1975), Pfaff (1979), Joshi
(1985) and notably Poplack (1980/2000) and Sankoff and Poplack (1981) to
propose that there must be constraints of a universal nature on where switching
could occur. Only brief examples of these constraints are given here. The clitic constraint The clitic constraint states that clitic subject or
object pronouns must be realized in the same language as the verb (Timm,
1975:478; Pfaff, 1979: 303). Thus, examples such as such as the following –
recorded in Strasbourg – would be marked with an asterisk as violations:
Example 1: French–Alsatian
il koch güet2
he cooks well
(Gardner-Chloros, 1991:168) The free morpheme constraint Poplack’s (1980) analysis of a corpus
collected in the New York Puerto-Rican community led her to propose that two
constraints were operating, the free morpheme constraint and the equivalence
constraint. These appeared simple enough to be universally applicable and have
il is a clitic pronoun in French according to Jones (1996).
been widely discussed (Clyne, 1987; Myers-Scotton, 1993b; Jacobson, 1998b).
The free morpheme constraint stated that a switch is prohibited from occurring
between a bound morpheme and lexical form unless the latter has been phonologically integrated into the language of the former (so a form like catch-eando,
with an English root and a Spanish ending is not a “permissible” code-switch).
As Bhatia and Ritchie among others point out, such coinages do in fact occur,
in both agglutinative and non-agglutinative languages. In some communities
this is one of the commonest forms of switching (cf. Eliasson, 1989, on
Maori–English). Here four different Swahili bound morphemes are affixed to
the verb ‘spoil’:
Example 2
vile vitu zake zi-me-spoil-iw-a
those things her they-perf-spoil-pass
Those things of hers were spoiled
(Myers-Scotton, 1983, quoted by Bhatia and Ritchie, 1996:640) The equivalence constraint Lipski (1978), Pfaff (1979) and Poplack
(1980) all formulated constraints stating that CS cannot occur at points in the
sentence where the surface structures of the two languages differ. Again, this is
undermined by examples such as:
Example 3: Greek Cypriot Dialect–English
Irthe dhaskala private
Came teacher private
a private teacher came
(Aaho, 1999:43)
English adjective–noun order is violated here. Note that English word order
would also be acceptable in Greek, yet the speaker chooses the alternative order
which is not common to both languages.
The notion of equivalence is itself problematic. Romaine (1989:118) remarked
that the equivalence constraint assumes that the two languages in contact share
the same categories and does not make predictions for language combinations
where this is not the case. Muysken (1995) also discusses the problem as one of
equivalence of structures in the two languages. Sebba (1998) goes further and
claims that equivalence is constructed by speakers rather than inherent in the
languages themselves.
None of the constraints proposed stood up when new language combinations
or new communities were studied (Agnihotri, 1987; Bentahila and Davies, 1991;
Clyne, 1987; Eliasson, 1989; Gardner-Chloros, 1991; Muysken, 1995; Nortier,
1990; Romaine, 1989, etc.). Instances of CS have been found in every grammatical position, not only by looking across corpora but even within a single corpus
(Nortier, 1990). The constraints have therefore come to be seen as having a relative
rather than a universal value (Romaine, 1995; Jacobson, 1998b; Muysken, 2000).
Grammatical aspects of code-switching
Poplack’s response to such counter-examples was to claim that apparent
violations were due to the fact that we were not dealing with CS, but with a
different process, “borrowing”. In Chapter 2, it was suggested that borrowing
affects all aspects of language, and can only ultimately be distinguished from
CS in diachronic terms. Poplack, however, claimed that borrowing could be
a once-off process (“nonce loan”) (see Poplack and Sankoff, 1984). The circularity of this argument dealt the model what many considered to be a fatal blow:
“Poplack’s defence of the structural integrity of linguistic systems is motivated
less by the evidence than by the desire to justify the validity of a particular
theoretical model of code-switching” (Romaine, 1989:286). Poplack (2000)
describes some relatively minor changes to the original proposals emerging
from later work: the fact that phonological adaptation is now recognized as an
unreliable indicator that a word is borrowed, and the formalization of the equivalence constraint on the basis that it is a consequence of speakers’ production
of “hierarchically and linearly coherent sentences” (Sankoff and Mainville, 1986;
Sankoff, 1998). But the basic circularity of the argument is unaltered.
Finally, little attention has been paid to the interaction of the proposed
constraints, and the possibility that they may not always be compatible with
one another, or may lead to different outcomes. Maters (1979) pointed out that
violating the free morpheme constraint by adapting CS words morphologically
repairs the ungrammaticality caused by violating the equivalence constraint:
Example 4
Tu peux me pick-up-er?
You can me pick up-INF suffix
Can you pick me up?
(Gardner-Chloros, unpublished example).
The speaker here “gets away with” violating the equivalence constraint –
switching from French to English at a point where the pronoun object placement
differs between the two languages – by giving the verb ‘to pick up’ a French
infinitival ending, -er. This sounds more “grammatical” in French than if she
had used the bare form. Such tactics should be of particular interest because,
rather like “flagging”, they indicate speakers’ awareness of grammatical difficulties such as equivalence – and, simultaneously, their determination to override them.
Generativist approaches Government Attempts to explain constraints on CS in terms of
Government typically contended that there can be no switching between a
governor and the governed element (Di Sciullo et al., 1986). This fails, however, to account for many common switches, such as those between verb and
adverb (Uno no podía comer carne every day, ‘We couldn’t eat meat every
day’), or subject NP and main verb (Les canadiens scrivono c, ‘The Canadians
write c’) (examples quoted in Muysken, 1995).
The proposals were modified in Muysken (1990) and restricted to lexical
government by non-function words. Even this prediction was too strong.
Muysken refers in particular to the numerous counter-examples in Nortier (1990)
from Dutch–Moroccan Arabic CS. These include switches between elements
canonically related by “Government” such as verbs and direct objects (anakandir intercultureel werk, ‘I I-am doing intercultural work’) or between direct and
indirect objects (ib li-ya een glas water of so, ‘Get for-me a glass of water or so’).
The theoretical constructs involved here are, in many cases, highly abstract,
and subject to frequent redefinition. In the case of “Government” for example,
several successive formulations of the relationship and the domain in which it
applies appear in the literature, and the class of governing categories has been the
subject of controversy within GB theory. The notion of “Government” has now
been abandoned in the Minimalist Program, partly as result of these definitional
difficulties (see Epstein, Groat, Kawashima and Kitahara, 1998: Chapter 1). “Null” theories MacSwan (1999; 2000) is one of the main proponents
of applying the Minimalist Program, as a formal model of grammar, to the
analysis of CS. The Minimalist model (Chomsky, 1995) is intended to be a
model of grammatical competence, and hence does not address aspects of CS
grammar which are likely to involve surface factors, or “performance” factors
such as processing and production (see section 5.2.2. above).
MacSwan (2000) assumes that the notion of “a language” should play no role
in the formal system employed to account for the data under analysis. The focus,
then, is on formal systems. Under the assumptions of the Minimalist Program,
the mixing of grammars is effectively the mixing (or “union”) of two lexicons,
as the significant features of grammars, including the parameters of variation
between grammars, are assumed to be located in the lexicon. Regarding CS,
MacSwan’s position is that “nothing constrains codeswitching apart from the
requirements of the mixed grammars” and that the grammar of CS will generate
all of the well-formed expressions which invoke elements contributed by more
than one language, and none of the ungrammatical ones (2005:2). In order to
identify ungrammatical expressions in CS, he uses some experimental data
(invented examples tested against informants’ intuitive judgements), as well
as naturally occurring data. However, this raises the problem of which “native
speaker’s competence” to appeal to, since this is an even more complex and
unreliable concept in the case of bilinguals than in the case of monolinguals.
Colina and MacSwan (2005) claim, like Poplack did, that phonological integration can help us distinguish between code-switches and (nonce) loans,
and thus that in Spanish–English code-switched speech tipear (‘to type’) is a
code-switch and typear [taipear] a (nonce) loan.
Grammatical aspects of code-switching
“Null” theories of CS grammar had been proposed by others before, e.g.
by Woolford (1983), Pandit (1990) and Belazi, Rubin and Toribio (1994). All
specify that CS can be described in terms of the grammatical principles relevant
to monolingual grammars, without postulating additional CS-specific devices
or constraints. Along with MacSwan, Mahootian (1993) and Chan (1999) have
continued this tradition. Mahootian’s proposals are couched in the formalism of
Tree Adjoining Grammar (Joshi, 1985), while Chan, like MacSwan, assumes a
Minimalist version of Principles and Parameters theory. Mahootian’s proposal
is mainly concerned with the content of lexical constituents, as determined by
language-specific rules for those constituents. Her proposal – which is essentially concerned with “surface” word order differences between languages – is
however called into question by counter-examples such as that below.
Example 5: German–English
Jemand hat gesagt dass er ist the father of her child
Somebody has said that he is the father of her child
(Eppler, 1999:287)
As Eppler observes, word-order constraints formulated in terms of the ordering
of constituents dominated by a specific node would not admit switches such as
this, in which the complement the father of her child follows ist, rather than
preceding it as the relevant phrase structure rule for German would require.
By contrast, Chan (1999) argues that certain patterns of switching can be
explained by reference to the types of phrase that “functional” categories
(Tense, Determiners and Complementisers, among others) select as their complements in different languages. He cites the example below from Bentahila
and Davis (1983):
Example 6: French–Moroccan Arabic
je peux le dire had le truc hada bas je commence à apprendre
I can it say this the thing this that I begin to learn
I can say this in order that I begin to learn
(Bentahila and Davies, 1983:323)
What is important here is that the Moroccan Arabic complementiser bas must be
followed by a finite clause in Arabic. Chan points out that although the complement clause is in another language – in this case French – the syntactic requirement that the subordinate clause be finite holds. Chan proposes that certain
instances of CS are constrained by the “Functional Head Constraint”, a condition to the effect that a switch can take place between a functional head in one
language, and its complement in the other language, provided that the complement matches the type of complement which would be required in the first
language. He claims that this constraint is empirically superior to similar
proposals by Belazi, Rubin and Toribio (1994), and Myers-Scotton (1993b),
in predicting a wide range of data without the need for special filters and let-out
However Chan’s analysis leaves several questions open. Firstly, Chan’s
constraint relates only to a particular set of categories. These categories (the
functional categories), as formulated in Chan’s theoretical framework, are,
as with Government, abstract categories, whose properties are not fully understood, and which do not constitute a homogeneous class. It is not clear, for
example, why functional categories should impose constraints on CS. An
alternative explanation of the example from Bentahila and Davis cited above,
for example, might be that verbs such as say (or its French equivalent dire)
require finite complements in most languages. Second, data from a range of
sources suggest that some “functional categories” such as agreement may
be affected in CS (witness the common phenomenon of the use of “bare” verb
forms in CS). It is by no means certain that the specific grammatical properties
of these categories are the same across languages. Nor is it clear that such
categories would consistently impose “constraints” on the form of switched
utterances. Chan’s Cantonese data, for example, contain aspectual markers
and these are argued to determine the absence of inflectional morphology on
English verbs which have been embedded in Cantonese sentences. In addition
to the problem of possible non-equivalence between the relevant Cantonese and
English functional categories (Cantonese appears to have no functional categories marking tense, for example), it is interesting to speculate how the situation
would work in reverse, say when an English verb appears fully inflected inside a
Cantonese sentence. Indeed, Chan offers examples of inflected English nouns in
Cantonese sentences. To sum up, while Chan’s analysis points to a potentially
interesting locus of research, it leaves certain questions unanswered. As with
other proposed syntactic “constraints”, the jury must therefore remain out on
these proposals.
The Matrix Language Frame (MLF) model
Klavans (1985) and Joshi (1985) among others, posited that CS involved a
“frame” or “matrix” into which elements of another language could be embedded. In a series of publications, Myers-Scotton and her colleagues have formulated an elaborate grammatical model based around this concept (1993b.). The
MLF (Matrix Language Frame) differs from earlier constraints-based explanations in providing a hierarchical framework and in tying in the proposed
constraints with:
(1) the role of different types of morpheme in CS. The basic proposal is that in
all CS there is a dominant language, the “base” or Matrix Language (ML).
This supplies the system morphemes (closed-class items) in the sentence,
whereas the embedded language (EL) supplies a proportion of the content
morphemes (open-class items);
(2) the psycholinguistic notion of activation (see Chapter 6).
Grammatical aspects of code-switching
“Grammar” as used here does not correspond clearly with any of the types
of grammar mentioned above. Language processing is said to involve the construction of a frame, generally dictated by one of the two languages (the ML),
into which elements of the other language (the “embedded language” or EL) are
slotted. The model is insertional, as opposed to the alternational approach
embodied by constraints (Muysken, 2000). The ML, we are told, is to be thought
of not as a language in itself, but as the “abstract grammatical frame of a
bilingual CP”.3 In certain cases, the ML may be “composite”, i.e. made up of
an “abstract frame composed of grammatical projections from more than
one variety” (Myers-Scotton, 2002a:22).
In her earlier work, Myers-Scotton had referred to the existence of “switching as an unmarked choice” (1983) – similar to Gafaranga and Torras’s “mixed
medium” (2001). But what is posited is a composite framework at the grammatical level. She claims that this occurs in situations of language attrition
or shift, when speakers do not have “full access to the morphosyntactic frame
of the ML”, or when “the notion of a target Matrix Language is not clear to the
speakers themselves”. This is a puzzling statement: it appears to imply that
the use of a Matrix Language is connected to the individual’s education or
competence rather than being a necessary aspect of production, as is claimed
elsewhere. The Matrix Language Leaving to one side the – possibly selfcontradictory – notion of a composite Matrix Language, which is said to arise
in very specific circumstances, various problems surround the definition of
the ML (and consequently the EL). The quantitative criterion Originally, the ML was determined by a
quantitative criterion. It was said to contribute the greater number of morphemes
in a discourse sample consisting of more than one sentence (Myers-Scotton,
1993b; 2002a). By Myers-Scotton’s admission, the size of the necessary
sample was uncertain: “How large is ‘large enough’ is an unresolved issue”
(1993b:68). Many bilingual conversations, according to this criterion, could
be seen as changing ML several times. As Bentahila and Davies (1998:31)
asked: “Should an interaction containing four sentences dominated by one
language that are followed by two more sentences dominated by the other be
analyzed as having a single Matrix Language, calculated on overall morpheme
frequencies, or should one recognize a change of Matrix Language within
the interaction?” Examples such as the following illustrate the problem: the
languages are pretty well equally represented, and some morphemes cannot
The CP is defined as a clause with a complementizer, though both the complementizer and other
elements in the CP may be null (Myers-Scotton, 1997:220).
easily be assigned to a particular language; in, de, en and is are all unclear or
borderline between English and Dutch.
Example 7: Dutch–English
Ja, in de, in de big place is het a lot, nou ja, je kan’t, t’ is de same als hier. Je habt
Melbourne en de other places met de high flats and so. Dat heb je in Holland ook.
Maar’n, maar a lot of places nou (now), de same before we go. D’r is, we go to my sister
in Apeldoorn, en zi hef de same place noog.
(Clyne, 1987:754)
Second, what is the relevant unit within which to count morphemes? Auer
(1998b) illustrates in some detail the problems attendant on segmenting utterances into clauses (or CPs), and the corresponding outcomes in terms of
determining the base language. The quantitative criterion was therefore eventually abandoned in favour of other criteria. The morpheme-type criterion The ML was said to provide the
function words, except within EL “islands”. The class of EL islands is openended, so the question is how such islands are to be identified without circularity. According to Myers-Scotton, islands are triggered by the use of a lemma4
which does not exist in the ML (1993b). As such islands are basically inserted
chunks, more complex types of CS had to be accounted for separately (thus
giving rise to the notion of a composite ML).
The division between function and content words is itself problematic.
Muysken (2000) points out that there are at least four different criteria relevant
to this kind of classification in different languages; also, the distinction does not
operate in the same way across languages. In a later paper, Myers-Scotton’s
collaborator Jake concedes that “there is variation across languages in the assignment of particular lexical ‘concepts’ to content or system morpheme status”
The MLF model has been repeatedly amended, partly in response to such
criticisms. In Myers-Scotton and Jake (2000), a new “submodel for classifying
morphemes into four categories”, known as the 4-M model, was presented. This
relies on a sub-division of system morphemes into sub-categories which are said
to be directly related to, and differentially activated during, the process of
language production. Accordingly, it is predicted that these different types of
system morphemes will be differently treated in CS, and indeed in other types
of language contact and change. But in fact the only evidence that these
morphemes are the product of different mental processes consists in pointing
to instances where they are treated differently in different instances of CS.
Myers-Scotton takes over Levelt’s definition of a lemma as the “nonphonological part of an item’s
lexical information including semantic, syntactic, and sometimes aspects of morphological
information” (Myers-Scotton, 1993b:49, quoting Levelt, 1989:6).
Grammatical aspects of code-switching
No independent criterion for ascertaining their different status is proposed. As
Winford (2003:164–165) points out, the classification of system morphemes
into four categories is not placed in the context of the earlier criteria. Most
recently, a discussion has developed between the MLF proponents and proponents of Minimalism as to whether aspects of both theories could be combined
to deal with the grammar of CS (Jake, Myers-Scotton and Gross, 2002;
MacSwan, 2005). The Matrix Language proponents argue that the ML concept
is the bilingual realization of a “Uniform Structure Principle”, which requires
that each constituent be uniform. This principle applies to monolingual production as well but is, as it were, “invisible” within it (Jake et al., 2002:72). Other characteristics of the Matrix Language The Matrix Language is
also said to be identifiable in other ways:
Psycholinguistic: the ML is said to be more “activated” in the brain. However
it is again unclear how this can be independently demonstrated. Also, the
connection between the language which is more activated in the brain and
the grammatical frame of a sentence is not made explicit.
Social: the ML is said to represent the “unmarked choice” for conversations
of that type in the community. Auer (1998a) remarks that this criterion
presupposes a very uniform community where linguistic choices are highly
constrained. In many cases where there is no social pressure to use either
of the two varieties, the alternation is related to discourse factors, as seen in
Chapter 4 (e.g. Alfonzetti, 1998 on Italian and Sicilian). Further possible grammatical criteria Others have suggested that
the ML is determined by the language of the main verb (Klavans, 1985;
Treffers-Daller, 1990). But as Muysken (1995) points out, many languages
have strategies to incorporate alien verbs (e.g. through prefixes in Swahili),
and taking the borrowed verb as determining the base language can be misleading. Further possibilities include Doron’s (1983) and Joshi’s (1985) claim that the
ML is determined by the language of the first major constituent in the sentence,
and Nishimura’s (1997) suggestion that word order is the relevant criterion. Use of the Matrix Language by other researchers A different
approach to the base language is adopted by Nortier (1990:158), who distinguishes between the base language of a whole conversation, and the Matrix
Language of individual sentences. Moyer (1998) contrasts the base language,
meaning the language which determines the grammar of the sentence, and
the main language, which “sets the frame for the entire exchange”. The latter
“can only be determined by taking into account the wider linguistic context of
the conversation or speech event” (p. 223). As Moyer suggests, we are not
dealing with a unitary phenomenon: our view of which language dominates
will depend on the level of planning – and the size of the corpus – which we
have to examine.
The MLF in its various incarnations has continued to be used as a convenient
means of describing large data-sets, sometimes despite evidence that it fails to
explain aspects of the data. For example, Rindler-Schjerve (1998) illustrated
what is described as a change of ML among the younger generation in Sardinia,
which is symptomatic of language shift. Most of what is said can clearly be
identified as Sardinian, and in over 93 percent of switches the speakers return to
Sardinian after a segment in Italian, though with considerable variations based
on age and network. At a grammatical level, however, instances of CS which
contradict the MLF model are found:
Example 8: Sardinian–Italian
No, deo, dae sos polcheddos suoso non happo tentu mancu … ca sa veridade, happo
accontentadu su cliente
No, I didn’t even make any profits from the pigs … the truth is, I satisfied the
If there is a Matrix here, it is clearly Sardinian. But the obligatory prepositional
accusative of Sardinian, which does not exist in Italian, is not realized in
the final part of the sentence (Italian: accontentato il cliente; Sardinian: accontentadu a su cliente). The EL has imposed its grammar on the ML – which the
model predicts as impossible.
Deuchar (2006), along with many others (Myers-Scotton and Jake, 2000)
found her corpus (Welsh–English) to be “overwhelmingly compatible” with the
various principles underlying the MLF. The framework deals best with cases
of very asymmetric bilingualism such as that which formed the basis of this
study. But, as Belazi, Rubin and Toribio (1994) had cautioned, different types
and levels of bilingual competence lead to divergent code-switching behaviour:
“The distinction between matrix and embedded language that is crucial to much
code-switching work may be indicative of a lack of balance in the two languages in the bilinguals whose speech is at issue” (p. 222). Muysken’s Bilingual
Speech (2000) offers an integrated framework showing that not only the MLF
but each of the grammatical frameworks we have considered offers only a
partial description of the grammatical possibilities in CS speech.
Muysken’s “bilingual speech”
Muysken prefers the term code-mixing (CM) to the commoner CS, reserving the
latter as a synonym for alternation, and distinguishes the following types: Alternation Alternation occurs when there is compatibility of the two
grammars, or at least equivalence at the point where the switch occurs. Models
such as Poplack’s, in which grammatical equivalence is a precondition for
Grammatical aspects of code-switching
switching, are seen as resulting from the fact that her Spanish–English data are
mainly of the alternational variety. Alternation is illustrated in data-sets which
vary considerably as to the patterns exhibited, but which share the feature that the
elements following and preceding the switched string are not structurally related. Insertion The second type is insertion, a process akin to borrowing
but where elements longer than a single word may be inserted. Muysken
suggests that the MLF model is directly related to the primacy of insertional
material in Myers-Scotton’s African corpus. Although no single criterion is
generally valid for establishing which language is the Matrix, in insertional CM
one language remains more activated and tends to provide the language of the
main verb and most of the functional elements. Models based on Government
represent a particular interpretation of insertion. Congruent lexicalization The third process is congruent lexicalization, in which the languages share a grammatical structure but the vocabulary
comes from two or more languages. Counter-examples to the constraints and
the MLF models, from data such as Clyne’s (1987) Dutch–English in Australia,
are considered instances of congruent lexicalization. The latter results from
grammatical convergence or from similarities between languages. Understanding
this type of CM involves carefully determining the nature of the monolingual
varieties which are mixed. Arguably this should be as much of a priority for the
other processes as well. Sociolinguistic factors determining type of code-switching Each of
the three types of CM is associated with different linguistic, sociolinguistic and
psycholinguistic factors. Alternation is likely to occur in stable bilingual communities with a tradition of language separation, each language being successively activated in the bilingual’s brain. Insertion is likely to be found
in situations where bilingual proficiency is asymmetric (e.g. colonial or recent
migrant settings). This is illustrated for example by Backus’s work on Turkish
speakers in the Netherlands (1999). Inter-generational language shift may be
reflected in a change in the direction of the insertion. Congruent lexicalization is
likely to occur between closely related languages, where their relative prestige
is roughly equal, or where there is no tradition of overt language separation (e.g.
second-generation migrant groups, post-creole continua); here the languages
are assumed to partly share their processing systems.
Muysken schematizes the three main types of CM which he identifies in a
triangle (see Figure 5.1) and shows how corpora collected in different settings
by Myers-Scotton (Swahili), Pfaff (Mexican American in Texas), Poplack
(New York and Ottawa) and Gardner-Chloros (Strasbourg) can be localized
within this space.
Mexican American
Puerto Rican
New York
Figure 5.1 Localization of a number of contact settings in the triangle
alternation, insertion, congruent lexicalization.
The importance of Muysken’s proposed correlations between types of CS
and sociolinguistic circumstances is that they provide us with a set of potentially
testable hypotheses. Although it is legitimate to describe CM in terms of the
grammatical regularities which characterize it in a given context, overall we are
faced with a degree of variation which no single set of grammatical rules can
account for. Muysken argues convincingly that both “constraints” and the
Matrix Language are relevant to certain types of CS only. This brings us back
to the argument that grammar can only provide one part of the explanation as
to why code-switching takes the form it does.
Is grammar the right framework?
Muysken (2000) remarks that code-switchers break various kinds of “rules” in CS
if those rules get in the way of their combining what they want to combine: “If you
cannot have equivalence, adopt another form, by bending the rules of the systems a
bit” (p. 32). This reinforces the idea that we need to look outside language as a
closed system, which has been the focus of research in the structuralist tradition
since Saussure and Chomsky, just as pragmaticians and discourse analysts have
done so successfully in the last few decades (Nehrlich and Clarke, 1994).
Code-switching outside the grammatical confines
For example, grammatical explanations have little to offer in respect of the CS
variously described as ragged (Hasselmo, 1972), paratactic (Muysken, 1995)
Grammatical aspects of code-switching
and disjointed (Gardner-Chloros, 1991), which is found in many contexts. This
involves speakers using pauses, interruptions, “left-right-dislocation” and other
devices to neutralize any awkwardness associated with CS.
Example 9: French–Brussels Dutch
Les étrangers, ze hebben geen geld, hè?
The foreigners, they have no money, huh?
(Treffers-Daller, 1994:207)
A better understanding of this may be found in “the grammar of speech” which
provides an alternative way to understand phenomena which formal linguists
would explain as a disparity between competence and – often imperfect –
performance. Biber, Johansson, Leech, Conrad and Finegan (1999) list various
characteristics of spoken English, based on a 40-million-word corpus covering
four different registers. They point out that over a third of all the units identified
are “non-clausal” and have an average length of two words. Production is “online”, showing the effects of limited planning time and being characterized
by “dysfluency, error, hesitations, repairs, grammatically incomplete utterances
and syntactic blends” (p. 1038). Such occurrences can “legitimize” combinations of typologically different languages, for example as regards word order.
Aitchison (2000) has pointed out that devices such as left-dislocation (see also
Geluykens, 1992), can find their way to being fully grammaticalized parts of the
language over a period of time.
“Flagged” switches fulfil a similar function. Such strategies reinforce the
notion that grammatical rules are present in the mind of code-switchers – albeit
sometimes in the negative sense of being something to be skirted round. A
suggestion for future research is to work backwards from examples where codeswitchers make use of paratactic strategies to try and see which grammatical
difficulties drive them to use such techniques.
Code-switching’s “language-like” properties
Boeschoten wrote that: “CS as verbal behaviour has language-like properties,
i.e., it is really not assumed to consist just of the combination of two completely
independent systems” (1998:21). This is an important remark, as specific
examples of innovation noted in the literature show. Bilingual compound verbs The creation of bilingual compound
verbs, as exemplified in 3.3.2, is so widely attested that it may constitute
a universal of CS (Muysken, 2000). In Edwards and Gardner–Chloros
(2007), these verbs are discussed as an example of bilingual creativity, which
highlights the inadequacy of describing CS as the sum of two pre-existing
Code-switching Compromise forms Where the languages are related, similar or identical sounding words, or “homophonous diamorphs” (Muysken, 2000), may
serve as a “bridge” facilitating the transition to the other language. As we saw
in example 7, function words such as de/the and dat/that, or in operate in this way
(Clyne, 1972). In a substantial corpus of Dutch–English and German–English CS
in Australia (Clyne, 1987), CS often coincided with the use of compromise forms,
making it difficult to establish with certainty where exactly the language switch
had occurred. Clyne also noted that Vietnamese– and Chinese–English bilinguals
sometimes switched unintentionally into English following the use of a word
corresponding to an English pitch (Clyne, 2003). Further “bridge” phenomena The use of compromise forms serves
to reduce the distance between two varieties. These include not only the “bridge”
words described by Clyne (2003) such as those exemplified above, but also bare
Example 10: French–Alsatian
Ah voila, nitt dass se do cueillir, un gehn dann uf d’ander Sit
Yes there you are, they shouldn’t pick, and then go to the other side
(Gardner-Chloros, 1991:159)
The French verb cueillir is an infinitive – it should be conjugated in the third
person plural in order to be grammatical in either French (cueillent) or Alsatian
in this sentence. But the Alsatian infinitive ending -iere (e.g. marschiere, to
march), is also the third person plural ending. The French infinitive is therefore
a compromise form, as the French infinitive ending -ir sounds like an Alsatian
conjugated third person plural. Double morphology Double morphology is a further “compromise
strategy”, whereby morphological marking from both participating varieties is
applied to the same words:
Example 11: Turkish–Dutch
Hollandaca ders Verdi
Pole-PL-PL-DAT Dutch [people] lesson give/PRET/3S
he taught Dutch to Poles
(Backus, 1992:90)
Myers-Scotton and Jake (2000) claim that the doubling up of certain types of
morpheme and apparently not of others confirms the distinctions between
different types of system morpheme which they draw in the “4-M” model.
Woolard (1999) argues that such “bivalent” elements, i.e. those which in
some sense belong simultaneously to both varieties, should be considered central
to an understanding of CS. Whereas they generally “drop out of the analytic
account” once they have been acknowledged, they in fact merit analysis in their
Grammatical aspects of code-switching
own right as a “socially meaningful, potentially strategic form of language
choice” (p. 9).
Sociolinguistic v. typological factors
If, as Muysken has argued, different types of CS are found in the context of
(a) different degrees of linguistic closeness between the languages and (b) different sociolinguistic circumstances, then it is a priority to find a way of assessing
the relative contribution of these factors. The obvious way to do this is to make
comparisons between cases where the same pairs of languages are combined in
different sociolinguistic settings, and different pairs are combined in similar
settings. We could then answer questions such as: How do the two aspects relate
to one another? Are the restrictions imposed by grammar the inescapable bottom
line, with sociolinguistic parameters merely pushing the patterns towards one set
of options rather than another? Or are the social, personal and interactional
reasons for CS the primary determinant, grammatical options serving merely
as second-order expressions of socially/individually determined choices?
Some comparative work was described in section 3.6.2 but a lot more remains
to be done.
Gumperz (1964) described a village in India where, through sustained CS,
two widely differing languages had all but merged at the grammatical level,
lexical differences alone being preserved. Later, in a study in Norway with
Blom, at the opposite extreme, two closely related varieties – two dialects of
Norwegian – remained functionally distinguished in the community’s usage
(Blom and Gumperz, 1972).5 Although Gumperz did not explicitly compare the
two cases, such contrasting findings confirm that the degree of typological
difference between the two languages involved in CS is not the only factor
involved in the linguistic outcome.
Other relevant comparisons have been made. Bentahila and Davies (1983;
1991) showed how different generations in Morocco, educated to differing
extents through the medium of French and Arabic, all code-switch but use
different proportions of the two languages and combine them in different ways.
In the Tyneside Chinese-speaking community, Li Wei (1998a) shows how
network patterns affect Chinese–English CS, as did Schmidt (1985) as regards
a declining aboriginal language, Dyirbal, and English.
Conversely, we find similarities in code-switching patterns, across different
language-pairs, where similar social circumstances obtain: for example amongst
close-knit groups of immigrants, CS is often not only very frequent but also
Since then, criticisms have arisen to the effect that the two varieties were in fact on much more of a
continuum linguistically than Blom and Gumperz implied (Maehlum, 1990). This, however, is
separate from the point at issue.
intricate at a grammatical level (Agnihotri, 1987; Cheshire and Gardner-Chloros,
1997; Nortier, 1990).
Third, in cases where there is CS between the same language-pairs in different sociolinguistic settings, the CS patterns can be radically different. In the
German–English data analysed by Eppler (1999), German SOV order is preserved in subordinate clauses, whereas in German–English CS in Australia, a
couple of generations down the line, the order shifts to that of English (SVO)
(Clyne, 1987:750–751). Basic word order appears to be relatively resistant to
change and is not “toppled” until a number of other symptoms of convergence –
or dominance of one variety over the other – have manifested themselves. This
too could form the basis of a future research hypothesis. Similarly, Muysken
(2000) has shown how the manner of incorporation of bilingual verbs varies
between Malay–Dutch spoken in Indonesia, and the same combination spoken
by Moluccans in the Netherlands.
He cites Huwaë (1992), in which Malay as spoken by Moluccans in the
Netherlands is compared with the results of earlier Dutch Malay contact
in Indonesia. The more intense contact between the two languages in the
Netherlands leads to a different way of integrating the same Dutch verbs
into Malay.
Example 12: Bahasa Indonesia/Moluccan Malay
Bahasa Indonesia
Moluccan Malay
Saya tidak me-reaksi a sing reageren
I NEG AG-reaction
I NEG react
I do not react
(Muysken, 2000:215)
Such findings contrast with language internal analyses which may lead to
incorrect predictions. One such prediction, based on typological considerations
alone, is that there could be no CS between languages with radically different
word orders (e.g Tamil SOV and Welsh VSO, Sankoff and Mainville, 1986).
While this particular contrast remains to be tested, there is in fact no evidence of
any such impossibility.
Typological difference is never the only factor determining the outcome of
CS. Treffers-Daller (1999) argues that the similar Romance/Germanic pairing
of French and Alsatian in Strasbourg and French and Brussels Dutch in Brussels
leads to similar patterns of CS despite sociolinguistic differences. The problem
is to decide how sociolinguistic differences (e.g. the different status of French
relative to the Germanic language in each case) can be weighed against linguistic similarities (see section 2.5).
Halmari hypothesizes that, sociolinguistic forces being equal, languages with
a rich morphology would tend to be Matrix Languages since they “bring their
inflections along with them” (1997:211). Johanson (2002), on the other hand,
points out that typological factors are not decisive in cross-linguistic contacts,
Grammatical aspects of code-switching
viz the multiple influences of Turkic on its differently structured neighbours,
and vice-versa. It should be a priority for future research to make systematic
comparisons between CS in different settings, taking account of both sociolinguistic and typological factors (LIPPS, 2000).
Some unconventional proposals
The difficulties of finding a grammatical “formula” to encapsulate CS has led to
some unconventional proposals, which lie outside the traditional grammatical
“Stand alone” segments (Azuma, 1998)
In an attempt to account for some switches which appeared to violate the MLF,
Azuma (1998) suggested that certain chunks, defined as “any segment which
can meaningfully stand alone in the speaker’s mind”, may be switched. Such
chunks are accommodated within a new category, the “stand alone category”.
An extreme example is the word the being the only English word in a Japanese
sentence, the purpose of the switch being to convey the message that what
the refers to is unique and matchless. Rather than supporting the MLF model,
this type of example points instead to the inadequacy of explanations relying on
grammar rather than discourse.
Speaker awareness/education (Pandharipande, 1998)
Similarly, Pandharipande (1998) tries to defend the notion that there are
universal structural constraints on CS in the face of her own evidence, from
Marathi–English and Marathi–Sanskrit, which shows variation in the degree of
convergence within the two pairs, depending on the speakers’ awareness of
the structural differences involved. Her argument, that the strategies for keeping
the two languages separate are themselves universal, appears self-defeating,
since what is actually shown is that the speakers’ degree of education – i.e.
a sociolinguistic variable – is necessary to account for the patterns which
are recorded.
Clausal chunks (Backus)
Backus (1999; 2003) proposes a different rationalization of certain types of
insertional CS which occur within Matrix Language clauses. He suggests that
clausal “chunks”, or “multimorphemic lexical items”, which can be viewed
as one idiom, may be inserted in the matrix; sometimes such chunks may be
discontinuous, i.e. allow “fiddling” with the tenses and other grammatical
elements. These may include plural nouns, verb–object, adjective–noun, etc.
collocations, and represent an element which “blurs the distinction between
grammar and lexicon”.
A grammar for speech?
Brazil (1995) focuses on a stretch of speech which he calls the “increment”,
proposing this as a speech grammar alternative to the sentence.6 The increment
procedes from the initial state to a target state through simple chaining rules,
based on the speaker’s perception of what communicative need must be satisfied
at the time, without reference to the constituent-within-constituent arrangements
of sentence grammars. In order to provide a satisfactory explanation of CS, such
a model would, ideally, need to be articulated with the grammatical level, since
CS occurs within increments as well as between them.
The MLF itself did in one sense try to break out of the boundaries of grammar
by linking the proposed framework to psycholinguistic production and sociolinguistic factors such as markedness, or “rational choice”. But the grammatical
framework itself, as Muysken argued, accounts less well for some types of CS
than for others. Rather than making yet further amendments to it, we should
consider from a more general perspective the implications of its fundamental
premise, i.e. that one language is always grammatically dominant in CS – for
whom is this true? Second, we would need to know what such a statement
would “buy” us even if it could be proved.
Clyne remarked that: “It is a matter of doubt whether the notion of grammaticality can be applied at all to data as variable as that of code-switching”
(1987:744). However this does not mean that CS is grammatically arbitrary.
Muysken (2000) has suggested, for instance, that similar typologies often lead
to CS based on equivalence between structures, whereas conflicting typologies
(e.g. opposing word orders) lead to different tactics being employed and different linguistic outcomes (Sankoff, Poplack and Vanniarajan, 1990). Muysken
also points out that “The looser the syntagmatic relation is in a sentence, the
easier it is to switch” (1995:188).
The difficulty with most models of CS grammar – both alternational and
insertional – is that they seek to describe the data in terms of the interaction of
discrete systems. The notion of constraints and that of base or Matrix Language
Brazil remarks on the obligation “to offer some explanation of how the notion of a sentence has
come to hold such a prominent place in our thinking about language”. He speculates that this is
due to “scholarly concentration, for long periods of history, upon the written word” (1995:227).
Grammatical aspects of code-switching
both imply that these systems are equated with an external notion of what a
language is, i.e. a variety which is the common property of a community. This
means that a whole bundle of features, grammatical, lexical, phonological, etc.
are stored and employed as a self-contained unit. But CS is of interest precisely
because it can provide insights as to how individuals’ underlying linguistic
competence is actually organized, as opposed to how the “languages” which
they officially “speak” might, in theory, mesh together. “The phenomenon of CS
confronts researchers with the problem of distinguishing between the idea of a
language as the product of an individual’s (grammatical) competence and that
of a language as an externally defined, self-contained entity” (Le Page, 1989;
Le Page and Tabouret-Keller, 1985).
Furthermore, as Jacobson (1998b) points out, several of the main grammatical
theories of CS seek to show that universal rules govern the alternation which is
observed, differing only as to what those rules are. In so doing they risk losing
sight of the role of CS as a mechanism of language change (Winford, 2003).
Myers-Scotton’s “ML turnover hypothesis” is presented as if one generation
speaks A with elements of B, and the next speaks B with elements of A. The
“rules” – being universal – do not change, only the order/role of the languages.
Yet language change is a gradual process, affected simultaneously by external
and internal factors, and such a view cannot account for this gradualness, which
is synchronically identifiable as inconsistencies between and within the speakers’ output. If CS either “obeys” the proposed constraints or, more rarely, does
not, the “deviant” instances remain to all intents and purposes unexplained.
Theories which do attempt to integrate the process of change into their
explanations include Boeschoten (1998), who has shown that even within a
small immigrant community, CS is tied up with the emergence of new norms.
He shows that it can provide a mechanism for grammatical, as well as lexical
change, for example by encouraging a preference for structures which show a
high degree of congruence between the two varieties. He illustrates this in
relation to CS sentences taken from a corpus of second-generation Turkish
speakers in Germany (p. 19). Pfaff (1979) had also pointed out a tendency in
Spanish–English CS to avoid structural conflict through the increased frequency, in mixed speech, of less popular stylistic options.
The final outcome of language contact is affected by many factors and
is inherently unpredictable (Heine and Kuteva, 2005). It may lead to effective
convergence of two or more initially separate varieties (Gumperz and Wilson,
1971), but other forms of partial re-alignment, cross-influence or renewed
divergence are also documented in different settings. A productive goal for
future grammatical studies of CS would be to look at CS behaviour as essentially creative: to identify the grammatical difficulties which code-switchers
face within any given language combination and the means which they employ
to get round these difficulties – what Sebba (1998) has termed “conceptual
work”. By examining the techniques which code-switchers employ, we may
indeed come across some universal strategies, as may be the case with bilingual
compound verbs (Edwards and Gardner-Chloros 2007).
In Chapter 6, we will look at another possible source of universality, i.e.
aspects of mental processing and production. In this chapter we saw that
independent motivation for the MLF model and the existence of a base language
were claimed to be directly related to the notion of psycholinguistic activation
(Myers-Scotton, 1993b:46–47). In turn, as we will see, some of the psycholinguistic literature refers to the notion of base language, accepting it as part of
linguistic orthodoxy (Grosjean, 2000). It is to be hoped that the different subbranches of the discipline are not building on one another’s sand.
Box 6 Code-switching in religious contexts
CS is frequently found in religious contexts, when the language of holy texts
comes into contact with vernaculars. The following example is taken from
an Orthodox Jewish newsletter, Hakohol, distributed in 2000 to households
in North London.
Below are the non-English words in Latin characters, in the order they
appear, followed by the phrases in Hebrew, glossed and translated in brackets. Grammatical studies of CS often focus on whether CS is made more
difficult – or even impossible – by the differences between the grammars of
the languages involved. In this written example, CS occurs despite the
different direction of Hebrew and English script (left to right as opposed
to right to left).
Halochos – Jewish Laws
Torah – the Hebrew Bible
Chazon Ish – the prename of an important Hassidic rabbi and an authority in
religious law, Rabbi Abraham Kerlitz (1878–1953)
Rebbes – Rabbis
Shulchan Oruch – a codex of Jewish law written in the sixteenth century.
Literally, ‘The set table’.
Shabbos – the Jewish Sabbath, beginning on Friday before sundown
and ending on Saturday night after three stars are visible
Shekiyas Hachamo – sunset
Sof Zeman Kriyas Shema – the end of time (of day, permitted) for
reading the Shema. (Shema, literally, ‘hear’: a prayer, which begins
with “Hear, O Israel, the Lord our God, the Lord is one”)
Tefillah Betzibur – public prayer
Grammatical aspects of code-switching
brocho – blessing
Talmid Chochom – a wise/learned scholar
Tefillas Haderech – prayer for the journey (recited whenever anyone
travels outside a city, any long distance)
Mingham – custom
Birkas Hagomel – blessing of thanks for being delivered from danger
Yotze the brocho – fulfil the obligation of blessing
Shaaloh – a question on a religious matter, to a Rabbi/religious authority
our redeemer of justice (acronym)
arrival of
Soon in our days, Amen (acronym) and Jerusalem Zion and the consolation
(… and may we be privileged to witness the – arrival of our redeemer of
justice and the consolation of Zion and Jerusalem soon in our days, Amen).
in peace
and you shall bring us back
(… one includes – and you shall bring us back in peace - a section from the
and kingship
name (of God)
(… and some say the brocho without – the name of God and reference to
God’s kingship)
the eye
from vanished
(… that it is not considered – vanished from sight)
Important Halochos for the Holidays
With the approach of the holiday season, we would like to point out a
number of Halochos, which require particular attention when travelling
away from the home and community environment.
Holidays spent in a Torah atmosphere have great benefits for the
spiritual and physical well-being of a family. In addition to benefiting
from more exercise, fresh air and a break in the daily routine, it is a time
when parents can spend time together with their children in a more relaxed
atmosphere, enabling parents to have a better understanding of their children’s needs and to devote more time to them.
On the other hand, it is of course essential that every aspect of this
relaxation be spent as dictated by the Torah. There are many areas of
Halocho which apply in holiday circumstances, which we rarely come
across during our regular routine, and which we may therefore be
unfamiliar with, and require special care. As the Chazon Ish writes in a
famous letter, “My Rebbes have taught me that in every move a person
makes, he must consider the halochos of the four parts of the Shulchan
May we all be privileged to experience a spiritually and physically
invigorating summer and return to our regular activities, healthy in body
and soul, and may be be privileged to witness the ‫ביאת גוא״צ ונחמת ציון‬
‫ױרושלים בב״א‬.
Times of Shabbos etc.
Before leaving for one’s holiday, it is important to find out the exact
times of the beginning and end of Shabbos, Shekiyas Hachamo, Sof
Zeman Kriyas Shema, etc. for the place one is going to. If possible, one
should find out about possibilities and times of Tefillah Betzibur in the
It is customary to take leave and receive the brocho of a distinguished
Talmid Chochom before setting out on a journey (Orach Chayim
Chapter 11o, Be’er Heitev 10 and Mishnam Berurah 28).
According to most opinions, Tefillas Haderech should be recited after
leaving town (if one forgot, one can say it later on the journey). There are
however opinions that it should be said in the morning, before leaving;
each person should follow their own minhag in this matter (See Be’er
Heitev 9 and Mishna Berurah 27).
If one is planning to return the same day (i.e. before daybreak of the next
day) one includes ‫ ותחזירנו לשלום‬in Teffilas Haderech to include both
If one travels for a period of a few days, stopping overnight, Tefilas
Haderech must be recited every day when setting out on one’s journey,
after daybreak.
If one travels over sea by boat, Birkas Hagomel must be recited. There
are different opinions regarding reciting Birkas Hagomel after travelling
over land by plane and it is preferable to be yotzei the brocho by hearing it
from someone else. There is also some doubt after crossing the sea by
plane, and some say the brocho without ‫שם ומלכות‬.
When sending luggage, which includes meat, fish, wine, bread, or other
foodstuffs to a holiday destination (or anywhere else) with a non-Jew, it
must be sealed in such a way that it is not considered ‫נתעלם מן העין‬, (Yoreh
Deah Chapters 63 and 118). If in doubt, a Shaaloh should be asked.
Psycholinguistic approaches
A central argument in this book is that CS forces us to consider questions
about what a language is, and that such questions are in turn essential to our
understanding of CS. These questions are of interest at a psycholinguistic, as at
other levels: if we can understand the mechanisms of the processing, decoding,
storage, retrieval and production of linguistic material, we may come closer
to understanding how language relates to other cognitive faculties. If it can be
established that, with respect to any of these aspects, bilingual functioning is
in some material way different from that of monolinguals, then there may be a
basis for linking this difference to aspects of bilingual speech – for example, for
confirming or rejecting the idea that there is always an underlying “Matrix
Psycholinguists have accordingly been intrigued by this opportunity to gain
a better understanding of how languages are stored in the brain, as well as
production mechanisms, and there is a substantial tradition of psycholinguistic
research involving bilinguals.1 As CS is principally a spontaneous and informal
phenomenon, however, studying it in its naturally occurring state is largely
incompatible with standard psycholinguistic methodological approaches.
The discussion here of psycholinguistic work on bilingualism, from a nonspecialist viewpoint, is based on the idea that CS should be considered from a
multidisciplinary perspective. The selection is based on those aspects which
have the clearest potential application to understanding CS. In most cases
the relevance to natural CS is still more or less indirect, though increasingly
the compatibility of theories with the facts of CS is used as a touchstone in
psycholinguistic work (e.g. Herdina and Jessner, 2002). There is still a reluctance
to straddle the sociolinguistic/psycholinguistic divide, and few cross-references
(with the notable exception of Myers-Scotton’s work) to the implications of
results in one field for theories prevalent in the other (though see Broersma
and de Bot below). For example, the parallel between proposals about the
As in other aspects of CS, the study of the specificity of individuals who speak more than two
languages is not yet highly developed, so most of the work here refers to bilinguals.
bilingual’s “language modes” (Grosjean, 1998, 2001) and the sociolinguistic
concepts of “audience design” (Bell, 1984), as well as accommodation
(Chapters 3, 7), has not, apparently, been explored. Yet looking at the same
phenomenon both from a psycholinguistic and a sociolinguistic/interactional
point of view could add considerably to our understanding.
The main impediment to such cross-fertilization is the methodological divide.
Psycholinguists generally avoid engaging with spontaneous, natural language,
preferring to use controlled, experimental data – language elicited in a laboratory,
measurable along as many dimensions as possible, and often simplified so as to
bring out the role of particular variables in the context of replicable studies. This
approach derives from a Western epistemological tradition which encompasses
all the major disciplines laying claim to the status of being “sciences”; a
discussion of the advantages and disadvantages of such an approach is beyond
our scope here.
The “scientific” study of language is of course essential. Psycholinguistic and
neurolinguistic research enhances our knowledge of the brain and may extend
our ability to treat various disorders effectively. But the demands of the clinical
sphere should not prevent communication between psycholinguistics and other
branches of the discipline.
Within its own terms, neuro/psycholinguistics seems to be making relatively
slow progress – as his parting shot to his 1999 Neurolinguistics of Bilingualism:
an introduction, Fabbro laments the poor state of knowledge of the “physiology
of language”. The evidence presented in this chapter suggests that some types of
progress might be faster if more account were taken of linguistic functioning in
naturalistic settings. Validity need not necessarily be sacrificed to reliability,2
and with some imagination, rigorous scientific method could be applied to “real”
CS data, collected in natural settings. The fact that this is rare for the time being
is reflected in the slightly different appearance of this chapter compared with
the others, in particular with respect to the small number of examples from codeswitched speech.
Some methodological issues
Forcing speech into “languages”
The most fundamental of these issues relates to the fact that psycholinguists
rarely appear to question whether and why each of the “languages” of bilingual
subjects should be seen as self-contained, discrete entities – in this, as we saw
in Chapter 5, they are in the company of other types of linguist, such as many of
In scientific terms, reliability refers broadly to a study’s replicability, and validity to its relevance
to the object of study.
Psycholinguistic approaches
those interested in grammar. Subjects of psycholinguistic studies, almost invariably, are said to speak, say, “English and Spanish”, or “French and German”,
without much consideration as to how the type and nature of their bilinguality
might influence the results. There is a dearth of psycholinguistic studies of
intermediate phenomena such as dialect or register-shifting, or of subjects who
speak an avowedly “mixed” variety such as a pidgin or a creole. Instead, most
studies focus on so-called “bilingual” subjects (usually university students)
whose mother tongue is a widely spoken and standardized language such as
English or Dutch, and who have learned another standard “language”, usually
limited to French or Spanish, in an academic context. The experiments often
involve testing them by requiring them to speak or react to these “languages”,
so as to investigate the underlying mental representation which these studies
purport to be exploring, thus leading to an unsatisfactory circularity in the
results. In Kamwangamalu and Li (1991), for example, Chinese–English bilinguals in Singapore had to decide whether ten sentences with Mandarin–English
switching were “English” or “Chinese”. This begs the question – why should
they be either? Furthermore, in natural circumstances such a decision would
normally never need to be taken, so the subjects were forced to make a decision
whose relevance to their actual behaviour is at best dubious.
Experimental tasks
The tasks given, as in the example above, often bear little relationship to
normal bilingual functioning. Azuma (1996) asked Japanese–English bilinguals
to switch languages whenever a particular sound was played, while they were
speaking “spontaneously”. Results indicated that in many cases, they chose
to maintain surface grammaticality by delaying the switch until later in the
sentence, or by repeating a chunk of what had already been said so as to avoid
switching in mid-flow. But their choices in no way resemble those which
underlie spontaneous code-switches, since all natural motivation for CS was
(a) lacking and (b) supplanted by an artificial one. To this we should add the
effect of being observed by academic researchers in an unnatural context, which
is hardly conducive to an output resembling their natural behaviour.
Toribio and Rubin (1996) report on a pilot study of CS in second language
learners (see Chapter 7). They asked English-speaking beginner, intermediate
and advanced learners of Spanish to repeat sentences containing CS, some of
which embodied what the experimenters considered to be “legal” code-switches,
and some of which contained “illegal” ones. “Legality” was determined by
whether the code-switches obeyed the Functional Head Constraint, a constraint
of UG, extrapolated to predict where switches “should” occur in bilingual
sentences. They found that the advanced learners had a tendency to turn
“ill-formed” into “well-formed” switches. The authors conclude that the less
competent learners were simply interpreting the CS sentences as English strings,
whereas the advanced learners were able to show UG effects distinct from those
of the first language. Such an experiment raises numerous questions, not least of
which is the fact that, as far as we know, none of these learners was a codeswitcher in their everyday life. Consequently, the advanced learners may simply
have been trying to demonstrate their competence by “correcting” the sentences
to conform to more traditional grammar.
Areas of psycholinguistic research with potential relevance
to code-switching
Psycholinguists divide up language functions into those concerned with (a) the
storage and representation of linguistic elements in the brain, (b) the neurological activation and inhibition which determine how these elements come
to the surface, and (c) production mechanisms (a sequence summarized in
the title of Levelt’s (1989) book, Speaking: from intention to articulation).
Historically, the majority of psycholinguistic studies, as of sociolinguistic
ones, were based on monolinguals. Studying bilinguals, however, brings to
the fore certain issues of storage and production. The study of bilingual aphasics, discussed below, is a good example: psycholinguists already derive substantial information about language and the brain from cases where monolingual
language does not operate smoothly (such as pauses, self-corrections and “tip
of the tongue” phenomena); the transitions and dysfluencies involved in bilingual production can similarly reveal aspects of these processes which would
otherwise remain hidden. To this one could add that as in other areas of
linguistics, one should study the bilingual majority rather than the monolingual
Separate and joint access to the bilingual lexicon
The mechanisms for keeping languages separate are one important focus. If
languages in the bilingual brain were totally separate, neither translation nor
CS would be possible; if they were totally integrated, then bilinguals would
presumably code-switch randomly all the time and would not be able to speak
monolingually. The question used to be asked in terms of whether there was one
“lexical store” in the bilingual brain or two. Much research has now shown that
bilinguals can never totally “switch off” one of their languages, and the question
is now asked in different terms: how does the bilingual “access” words from
their different languages appropriately for the task in hand? Whereas many
studies throw light on how this is done at the level of perception – since input
can be manipulated experimentally – very little is yet known about production,
since it cannot be manipulated in the same way (de Bot, 2004).
Psycholinguistic approaches
The concept of “language tags” is often used in models which attempt to
explain this dual ability (Albert and Obler, 1978).3 One of the features attached
to words in the brain is assumed to be the language to which they belong.
In such studies, the lexical level has pride of place, i.e. how individual words
are stored, connected with one another and with their meanings, and how they
dictate the grammatical structures within which they appear in speech. The
proposal that this level has primacy is known as the “lexical hypothesis”
(Levelt, 1989).
Intensive CS in normal subjects suggests that the varieties are closely interlinked in the brain, and simultaneously accessible. Decisions as to which words
to utter in which language appear to take little or no measurable time (though
see the discussion below about phonological aspects of CS).
Encoding and production
Second, psycholinguists have attempted to determine at what point in the
production process the encoding language is selected, which in turn pinpoints
the moment when the “decision” to code-switch becomes relevant. Meuter
(2005) suggests that the available evidence on activation indicates that all related
lexical representations are activated late in the selection process,4 and that there
are four main types of factor relevant to the selection of a language:
(1) global “de-selection” of the undesired language – though we cannot be sure
the de-selection is global or whether it only affects the relevant parts;
(2) the effect of proficiency: it has been shown in various experimental tasks
that it is more difficult to inhibit the L1 than the L2, but that once that has
been done switching back into the L1 takes longer than continuing in the L2
(as measured in milliseconds);
(3) the factors that trigger a switch;
(4) the monitoring capacity, which allows the selected language to be maintained.
These processes are obviously largely subconscious, as speakers are on the
whole not very aware of their code-switching behaviour, and tend to be surprised at their own performance if you play it back to them (see section
Since we cannot observe the language production process directly, various
models for it have been proposed (see below). Along with the evidence from
bilingual aphasia and studies of the localization of languages in the brain, these
Defined there as “a feature label associated with each individual item”.
Slips which arise during conference interpreting may provide a useful source of data.
Simultaneous interpreters, particularly when they are working “bilingually” – i.e. with two
languages, both used passively and actively – sometimes accidentally paraphrase in the original
language instead of translating into the target language. This may imply that the effort which goes
into understanding and generally processing the input takes precedence over the decision as to
which language to choose for the output. See also the question raised in section 6.6.1.
are the principal areas of psycholinguistics with a bearing on CS which will be
considered in the remainder of this chapter.
Bilingual aphasia
Since experimentation on human subjects is limited for ethical reasons, the
study of disorders of the language system has traditionally been used as a source
of information about the underlying representation of languages in the brain.
The recovery patterns of bi/plurilinguals who suffer brain lesions (“aphasias”)
and lose some of their language abilities as a consequence have been studied
and described for over a century (Hyltenstam and Stroud, 1989). By doing so,
researchers hope to find out more about the physical location of language
faculties and also about their functional interaction in tasks where the use or
knowledge of several languages is required (Clark and Clark, 1977).
A methodological caution
Grosjean (1989) has warned that before drawing conclusions from such subjects, it is essential to have some idea of their premorbid or baseline language
habits, otherwise one might, for example, assume that they were code-switching
as a result of pathology, when this was in fact their normal mode of speech.
Second, Grosjean pointed out the crucial role of specific methodological considerations in the testing of bilinguals, such as the need to use a monolingual
researcher when testing the performance of bilinguals in monolingual tasks
(2000), as a bilingual tester might bring forth responses affected by the subjects’
bilinguality. Alongside the question of in what sense the languages are independent entities, Grosjean’s admonitions point to experimental pitfalls which
may undermine the conclusions of research in this field.
Brain scanning techniques
In the future, brain scanning techniques such as CAT, PET and MRI5 may
provide further insights regarding storage, activation and production (see Vaid
and Hull 2002:343–350). Currently, subjects to whom these techniques are
applied are studied in a heavily controlled environment (sometimes placed
in a scanner or even wearing a special cap with electrodes), and the type of
“linguistic reactions” which are studied consist of brain activity in response
to single word stimuli (Green, 2001; Grosjean, Li, Münte and RodriguezFornells, 2003). Some of the conclusions drawn from these studies are, clearly,
CAT = Computerized Axial Tomography, PET = Positron Emission Tomography, MRI = Magnetic
Resonance Imagery.
Psycholinguistic approaches
informative about language control in bilinguals and potentially relevant to CS,
e.g. the finding that bilinguals find it very difficult to completely “turn off” their
response to either of their languages. There is also some evidence that linguistic
activity is represented in the same cortical areas in balanced bilinguals and in
different areas in less proficient ones (de Bot, 2004). However, we are still a
considerable way from being able to use these techniques to study spontaneous
Emotional factors
The literature also contains a large number of interesting case studies (e.g.
Fabbro, 1999; Grosjean et al., 2003; Vaid and Hull, 2002). For example, there
is growing evidence that emotional factors have a considerable impact in how
one learns, remembers and uses languages (Dewaele, 2004; Pavlenko, 2004;
2005). Switching to an L2 may serve a distancing function (the “L2 distancing
effect”) or allow the speaker to avoid anxiety-provoking material, whereas the
L1 elicits more personal involvement. Pavlenko (2005) cites the following
significant example from a psychotherapy session described by Rozensky and
Gomez. The switch from L2 to L1 leads the speaker to become more emotional
and to recognize feelings which were not otherwise articulated:
Example 1
p: I don’t want to do it. I really don’t want to go to the hospital. I’ve already been to
the hospital so many times for tests and everything else.
t: It sounds like your doctor thinks it’s important.
p. I know. I know. I suppose I should do it and get it over with but I don’t know …
t: Do you have any idea why you are so against the idea?
p: [Silence] I am not sure. I don’t know how to describe it.
p: Can you try?
p: It’s hard, it’s so hard.
t: Can you try it in Spanish?
p: [Tears starting] No quiero. Tengo, tengo miedo. [Silence]
[I don’t want to. I’m, I’m afraid]
t: Miedo de que?
[Afraid of what?]
p: [Crying hard] Que no voy a salir.
[That I’ll never come out].
(Rozensky and Gomez, 1983:156)
The different emotional roles of the bilingual’s languages are also reflected
in some studies of bilingual aphasia. One well-known case concerns a multilingual patient, who, following a stroke, was only able to talk in a language he
had not used for years, French, which had had great sentimental importance
in his life, as he had had a passionate love affair with a French woman in his
Example 2
The patient began to speak again two or three days after the fit, but to everyone’s surprise
he spoke in the beginning only French, first stammering out [a] few words,
then successively more and more correctly. His wife did not understand him, and his
children with their poor school French acted as translators between their parents. (Grosjean,
1982:229, reporting Minkowski, 1928)
The close connection, which these examples illustrate, between the languages spoken and emotions is likely to prove highly relevant to our understanding of the motivations behind CS. Pavlenko mentions that the effects of
switching to or from a mother tongue may also be at least partly to do with
pragmatic differences between the languages themselves: Guttfreund (1990)
reports on a series of psychological tests administered to Spanish–English bilinguals, of whom half were Spanish-dominant and half were English-dominant.
Both groups exhibited more anxiety and depression in Spanish, reflecting the fact
that showing affect in Spanish may be more acceptable than in English.
Network failures
Information gleaned from bilingual aphasia also has a bearing on the issue of
language separation v. conjoinedness. For example, aphasia can result in an
inability to switch into one of the subject’s languages, even though it is clear that
comprehension is preserved; alternatively, the ability to speak a language may be
preserved while comprehension is missing. It is even possible for both situations
to pertain at once: Albert and Obler (1978) report the case of a patient who had
Broca’s aphasia in Hebrew (i.e. she understood but could hardly speak it), and
Wernicke’s aphasia in English (i.e. she spoke but could not understand it). There
are also reports of stimulation of certain areas of the brain (e.g. during brain
surgery under local anaesthetic) giving rise to a flow of language, which sometimes involves CS when the patient is bilingual (Lebrun, 1991; Fabbro, 1999).
The speech of brain-disordered individuals, as one might expect, often shows
other signs of confusion along with unruly CS, as demonstrated by the patient
E.G., who suffered a stroke at fifty-five. He was of Slovene mother tongue, with
Italian as a second language, Friulian as a third, and English as a fourth.
Example 3
examiner: What was your job in Canada?
In Canada? Co facevo la via? I was working with ce faccio coi..del..fare,
i signori la che i faceva
In Canada? What I did there? I was working with I do with..do..men
there who did..
(Fabbro, 1999:154)
Other types of bilingual disorder which have a bearing on this issue include
“alternating antagonism”, in which patients are only able to express themselves
Psycholinguistic approaches
in one language at a given time; inability to translate between languages although
both are clearly understood; and, conversely, “automatic translation”, a compulsion to translate everything said in one language into another.
Perečman (1989) lists the types of pathological mixing encountered, including word mixing, root and suffix mixing, blending of syllables, use of the
intonation of one language with the lexicon of another, the syntax of one with
the lexicon of another or words from one language pronounced with the
phonemes of another. She proposed to correlate each of these types of mixing –
lexical–semantic, syntactic, morphological, phonological – with the production
stages outlined in Garrett (1980). These four stages concern:
(1) conceptual relations established at message level;
(2) selection of words relevant to the sentence, at functional level;
(3) closed-class words selected at positional level, and phonological form
(4) phonetic form of words specified at phonetic level.
The implication is that the different types of switching observed originate in
one of these levels of planning. This would seem to correspond well with
the fact that in normal CS, several entirely different types of CS can co-exist
in the same conversation and have quite different motivations. Giesbers
(1989), discussed in De Bot (1992/2000), similarly suggests that different
types of CS are linked to different modules in the language production
system. Amongst these, Giesbers identifies unintentional CS (“performance
switches”), not motivated by contextual factors or linguistic need but resulting
from interference – which he considers “bilingual slips of the tongue” (see
example (4)).
Other types of bilingual aphasic behaviour are distinct from CS as such,
e.g. the phenomenon of spontaneous translation or language “tagging” (in a
different sense to that used above – here it means naming the language of the
question instead of replying to the question) (Perecman, 1989). It is important to
note, however, that bilingual aphasics do not necessarily mix their languages in
a pathologogical fashion, and in some cases their CS shows the same kind of
grammatical patterns as that of non-disordered people for whom CS is an
everyday mode of speech (Hyltenstam and Stroud, 1989). Marty and
Grosjean (1998) found in an experiment that aphasic bilinguals retained the
ability to control the language “mode” in which they operated (see discussion of
this concept below), in other words to switch to a monolingual mode when
addressing a monolingual interlocutor.
The conclusion from the aphasia studies as far as CS is concerned must be
that the same sort of disturbances which affect linguistic abilities generally may
also affect CS, and not that any breakdown in language faculties results in CS.
Only in the most extreme cases is the CS of aphasic patients significantly
different from that of certain non-aphasic bilinguals. The studies therefore
support the idea, now prevalent in the sociolinguistic literature, that CS is “just a
normal way of speaking”.
Localization and lateralization of the bilingual’s
languages in the brain
The search for the localization of languages in the brain has a long history.
The major breakthrough with respect to where specific language functions are
localized came from the work of Broca (1861), who discovered that the left
hemisphere’s third frontal convolution was an important centre for language
production, and Wernicke (1874/1977), who identified various anatomical
pathways, as well as the memory area, for the sound and the motor images of
words. Since then, numerous different types of study have confirmed that for a
high proportion of right- as well as left-handed people, most language functions
are located in the left hemisphere of the brain (Obler, Zattose, Galloway and
Vaid, 2000; Fabbro, 1999).
Different languages, different locations?
Demonstrating that different languages are stored in different areas of the brain
proved much more difficult and controversial. While some believed this to be
the case, others argued strongly that the languages are kept apart by neural
mechanisms only (Paradis, 1981; Paradis and Goldblum, 1989). Real-life cases
are rarely sufficiently clear-cut to give unequivocal support to either position.
For example, although the brains of bilingual aphasics are frequently examined
post mortem in order to determine the locus of the damage, few patients
demonstrate one isolated language deficit (e.g. total loss of one language),
such that one could clearly correlate the lesion with the use of that language
and that language only.
Locating the switching mechanism
Fabbro (1999) does however mention Pőtzl’s (1925) observation, on the basis
of a brain-damaged patient who lost the ability to switch away from the
language spoken just prior to insult, that the ability to switch itself was located
in the left parietal lobe (see also Fabbro, Skrap and Aglioti, 2000). Meuter
points out the potential relevance of the fact that switching behaviour in general
(e.g. the ability to switch tasks) is controlled in the frontal lobes (2005:
363–364). Similar questions arise for classic linguistic CS and switching
between different signing systems in the deaf community, or switching between
a sign language and a vocal language where speech and signs are produced
simultaneously rather than sequentially (Davis, 1989; Ann, 2001).
Psycholinguistic approaches
Pőtzl’s observation ties in with a tradition of research in which it was assumed
that there was a specific on/off switching mechanism in the brain. In 1959,
Penfield and Roberts proposed a “single switch model”, which suggested that
the activation of one of the bilingual’s languages rendered the other inoperable.
There followed a series of papers on the “two-switch model”, according to
which there was an “output switch” for production and an “input switch” for
processing incoming language (e.g. Macnamara, 1971). It was proposed that the
time taken by the input switch could be measured by comparing the reading
times for monolingual texts and for comparable texts containing deliberate CS.
The contrived nature of this activity, however, cast doubt over the validity of the
conclusions. Neufeld (1976) rightly pointed out that there are different components involved in CS, phonological, lexical, semantic, etc., and that different
experimental tasks would call these into play to differing extents. Albert and
Obler (1978) argued that the control system of bilinguals is more continuous
and flexible than the single or dual switch models suggest, and that neither
language is ever totally excluded. This approach has been taken forward by
Grosjean and Green among others (see below).
One pattern of lateralization for all?
Albert and Obler (1978) claimed that linguistic functions in bilinguals are, more
often than in monolinguals, represented, to some extent at least, in the right
hemisphere of the brain. This is held to be the case particularly where the second
language is learnt later in life and by more explicit means. Paradis (2000) is,
however, strongly critical of psycholinguists’ endless revisiting of the lateralization question. He shows that for each study which has found the bilingual’s
languages to show different lateralization patterns, another can be found to
show that the pattern is the same as in monolinguals, or that no clear conclusion
can be drawn (Obler et al., 2000).
This brings us back to the point made earlier, that traditionally, psycholinguistic studies did not take much account of individual differences in bilinguality, though this is now changing (Herdina and Jessner, 2002; de Bot, 2004).
Where, exceptionally, some account was taken of these differences, such as the
age when subjects started learning their L2, interesting results were obtained.
For example, there appears to be a reasonable measure of agreement among
researchers that a second language learned after puberty is more likely to be
represented, at least to some extent, in the right hemisphere – whereas language
functions overall are represented in the left hemisphere. Children who suffer a
brain trauma in the left hemisphere before the age of around ten can “relocate”
language functions to the right; and in cases of an aphasic lesion affecting the left
hemisphere, it may be easier for a patient to recover an explicitly learned language
(even one which was never actively used, such as Latin or Ancient Greek), rather
than their mother tongue, which was acquired through unconscious mechanisms
and is represented in “implicit” memory systems (Fabbro, 1999).
Drawing conclusions from lateralization studies
Even if we were able to show clear, unequivocal patterns of lateralization or
localization of the language functions in bilinguals, it remains unclear what
exactly this would tell us about bilingual functioning. The fact that, in practice,
it remains so difficult to pin down the bilingual’s languages to different locations in the brain after numerous attempts, implies that bilingual competence is
on the whole a well-integrated capacity (particularly in the case of varieties
learnt simultaneously as co-mother tongues).
The difficulties of drawing unequivocal conclusions from clinical evidence
about the organization of languages in the brain suggests that “languages” as such
may be the wrong units to be looking at, and that we should be looking instead
at groupings of features, which vary from bilingual to bilingual, and are the
building blocks of languages. Sometimes, these blocks may be assembled in such
a way as to produce what we identify as a monolingual utterance – though even
then we should be aware that what appears to be a monolingual utterance to a
monolingual interlocutor may in fact be made up of elements from more than one
language. Each of the bilingual’s languages is informed and altered by their
competence in the other – even one which is no longer used or “remembered” at a
conscious level. The fact that such “forgotten” languages are still imprinted in the
brain is demonstrated by the fact that in some cases they have been reactivated
under altered states of consciousness such as hypnosis (Fabbro, 1999).
Models of bilingual production
Psycholinguists make extensive use of “models”, which are often schematized
in the form of diagrams in which the important factors are framed and connected
to one another by lines or arrows to show the causal/temporal/implicational
relationship between them (see Figure 6.1).
This method of presenting and systematizing theoretical proposals is much less
common in other areas of linguistics. As soon as one gets beyond a basic diagram
such as Figure 6.1, questions arise as to the exact relationship of the components
within the model to one another and also of the model overall to the behaviour
which it proposes to schematize. In particular, the type and extent of counterevidence which renders necessary a revision of such models is an open question.
Several such models have been developed to explain bilingual performance.
These have been discussed at length in the psycholinguistic literature and only
a highly selective account can be given here (e.g. Kroll, 1993; Fabbro, 1999;
Clyne, 2003; Dijkstra and van Hell, 2003). The models are focused on bilingual
Psycholinguistic approaches
Lexical Links
Conceptual Links
Figure 6.1 Hierarchical model of billingual memory representation (Kroll,
1993: p. 54)
production in general rather than CS, though attendant proposals as to how CS
can arise are a useful by-product. Both Clyne (2003) and Fabbro (1999) add
their own comments and suggestions – in Clyne’s case amounting to an alternative model (2003:213). The main models discussed here are Green (1986/
2000) and De Bot (1992/2000) (both reproduced in Li Wei, 2000). Clyne also
pays considerable attention to Dell (1986), Grosjean (1998) and Myers-Scotton
and Jake (1995). An influential older proposal to explain how lexical items are
represented in the brain, with potential applications to bilingual performance, is
Morton’s “logogens” (1979).6
As stated, an overall understanding of bilingual functioning is the aim, but
for present purposes, we will concentrate on the types of explanation which this
research offers for CS in particular.
An advantage of Green’s (1986/2000) model lies in its economy: it accounts
for both normal and pathological language behaviour in bilinguals, using the
notions of control, activation and resource. Activation is a complex psycholinguistic concept, which depends on such factors as the amount of contact with
and use of a language, the level of proficiency, the method of instruction, age of
acquisition, etc. De Bot (2004) provides an image to explain activation and its
opposite, inhibition, which he likens not to an on/off switch but to “holding
down ping-pong balls in a bucket of water” – where occasionally, however hard
one tries, some will pop to the surface.
Each of the bilingual’s languages, and each element within them, has an
“activation threshold” which depends on how often, and how recently, they
Logogens were said to be “devices that collect evidence for the appropriateness of a word”.
Information which activates logogens originates in the cognitive system. Each logogen has a
threshold. When the collected bits of evidence exceed the threshold, the logogen “fires”, i.e. is
ready for use.
Figure 6.2 An inhibitory model for a bilingual speaker within the control,
activation and resource framework (Green, 2000: Fig. 16.2)
Notes: →, flow of activation; ―▄, control instructions; →→, resource input;
― ― , inhibitory control.
have been used (Fabbro, 1999). The activation level for comprehension is lower
than that for production, which, as Green explains, is why some aphasics can still
understand a language, though not speak it (the same applies in normal subjects
when presented with a second language which has gone “rusty”) (see Figure 6.2).
However, Dijkstra and van Hell (2003) point to various questions inherent
in the “activation” metaphor. Does it imply that there is an active representation of the language as a whole? How can a language be activated without
actually being used? To what extent does activation depend on “top-down” –
i.e. contextual – factors?
The converse process, “inhibition”, is the process whereby, a given word
having been selected, the possible alternatives are suppressed, including, in
bilinguals, words in the other language. Similar processes in monolinguals
presumably suppress words from the wrong register or in the wrong dialect.
Experimental evidence suggests that bilinguals are only partially able to inhibit
the activation of one of their languages in selective word recognition tasks.
Dijkstra and van Hell therefore advocate the activation metaphor being applied
in a well-specified and testable way only.
Psycholinguistic approaches
Green writes specifically about CS (1986/2000:414), so we will look at what
he says in some detail:
In the case of code-switching, there need be no external suppression of L2 at all; at least
in the simplest case, such as continuous word association, the output can be free to vary
according to which words reach threshold first. Indeed in this circumstance, mixing
languages is certainly no slower than producing associations in only one language . . . In
the case of normal speech a word cannot be produced unless it fits the syntax of
the utterance. Accordingly, for example, an adverb will not be produced in a slot
requiring a noun. Switches then will obey the syntactic properties of the two languages
although no special device or grammar is required to achieve this goal. Code switches
most often involve single words, especially nouns …, though ones involving phrases or
entire clauses also occur. In these latter cases, we suppose that structures from L2 reach
threshold earlier. Since any words produced must meet the structural conditions, such a
scheme predicts that code switches will preserve the word order in both languages.
The second part of this passage, which has been italicized, raises certain problems if one looks at it in a wider – i.e. not purely psycholinguistic – framework.
First, in what sense is it true that a word cannot be produced unless it fits the
syntax of a sentence? This statement seems to stem from theoretical linguistics.
What is meant by the word cannot, since we know that this does happen? As
we have seen in Chapter 5, speakers do violate grammatical rules. The shifting
of a word from one syntactic category to another, also said to be excluded, is a
productive technique both for monolinguals and bilinguals.
The idea that CS requires no further grammatical apparatus beyond that
provided by the two existing languages (the “null” hypothesis) has been discussed in section It leaves out of account the creative and innovative
devices which particularly characterize CS (e.g. new operator verb constructions,
see Chapter 5). Why should the techniques for doing this not be considered
part of linguistic competence? The assertion that code-switches most often involve
single words is true in some communities but not in all, and cases have been
reported where inter-clause or inter-turn switching is more common (see section
4.4.1). The last sentence of the quotation is again derived from the grammatical
work on CS, in this case Poplack’s work on constraints (to which Green refers).
There are in fact numerous cases, as we have seen in Chapter 5, where CS does not
preserve the word order in both languages. The model is focused on the individual
only, and an idealized individual at that. It does not cater for the many variations in
CS behaviour which sociolinguists have demonstrated.
To what extent is Green’s model still helpful to our understanding of this
phenomenon? The idea that when speaking monolingually, bilinguals are inhibiting one of their languages is widely accepted, and accounts for involuntary
switches – failures of the inhibitory system. The system slips when the speaker is
tired, stressed, etc., as bilinguals will attest and psycholinguists have pointed out
(Javier and Marcos, 1989).
Example 4
An old lady speaking to a monolingual French-speaker in Strasbourg was recounting a
story, and in the middle of the story, which was told entirely in French, she slipped in the
phrase un dann hab’i g’saat.. (and then I said) in Alsatian before carrying on with her
(self-) quotation in French. (Gardner-Chloros, 1991)
Her inhibition system had failed for that short, framing clause, and she had
switched exactly as she would have done had she been speaking to a member of
her bilingual, code-switching network. Matras (2000b) suggests that connectives may be particularly subject to such non-volitional switching, and that this
in turn may lead to specific language changes.
The aspects of activation which are basically “intention-neutral” – e.g. based
on frequency of use, “availability” of particular words, etc. must be balanced
with the fact that much of CS has a clear function, discourse-related or otherwise
(Auer, 1998b). Green does not explain how “deliberate” activation relates to the
more mechanical kind, in which whichever word “reaches threshold first” is
used. What happens in case of conflict between the two types, i.e. when a word
comes more easily to mind in one language or is in the language required by
interlocutor-related factors, but is in competition with a more suitable equivalent in the alternative language? What determines the outcome of the battle
between conflicting activations? An answer may lie in the “specifier” in the
node above – a unit which specifies which language should be spoken and
whether the subject should code-switch or translate, i.e. operate between the two
systems – but so far the role of the specifier remains somewhat mysterious. The
problem is reminiscent of the conflict between languages as indexical of social
factors and languages as conversational and pragmatic tools (section 4.2.4), and
goes to the heart of the difficulties in analysing CS.
The final element of Green’s model, which was added to his earlier version,
is the notion of “resource”. This idea “makes explicit the fact that a system
needs energy to operate” (1986/2000:412) and, as can be seen in Figure 6.2,
the “resource generator” occupies a central position relating to word output in
each of the languages and putting in the necessary “energy” before the final
phonological assembly. The use of this concept allows a bridge to be made
between studies of normal and brain-disordered speakers: in various disorders
the language system may not be lost, but the patient may simply lack the
necessary resources to activate it. Recovery implies a renewal of the resources
which power activation and inhibition.
De Bot
De Bot (1992/2000) set out to make the minimum possible number of adaptations of a well-known model of monolingual production by Levelt (1989),
so as to make it adequate for bilingual production. Rather than offering a
Psycholinguistic approaches
discourse model,
situation knowledge,
encyclopedia, etc.
preverbal message
parsed speech
phonetic plan
(internal speech)
phonetic string
overt speech
Figure 6.3 Levelt’s speech production model (Levelt, 1989: Fig 1.1.)
diagram of his own, he reproduces Levelt’s Speech Production Model (see
Figure 6.3), and discusses in what ways it needs to be adapted. Supplementary
information is taken from De Bot and Schreuder (1993), which particularly
concerns CS.
The model includes a “Conceptualizer”, in which intended meaning is planned
and converted into “preverbal messages”; a “Formulator”, where this message
is converted into a “phonetic plan”: this is where the appropriate lexical units,
grammatical and phonological rules are selected; and an “Articulator” which
converts this plan into actual speech. At the Formulator stage, a distinction
between two aspects of a lexical entry is relevant: on the one hand it consists of
a “lemma”, representing the meaning and syntax of the lexical entry, and on the
other hand the “lexeme”, its morphological and phonological form.
Levelt’s proposals as to how speech goes through the various planning stages
before emerging on the surface is one of the main models in the field. Its
extension by De Bot to bilingual production increases its potential applications
and testability. However, what exactly is meant to be happening at each of these
stages is not transparent, and raises again the question of the exact status of such
models. As De Bot writes:
In order to try to clarify the workings of this model I will illustrate the various components
by using an example. Imagine we want to say: The train from Amsterdam arrives at
platform four. We know from our knowledge of the world that trains regularly arrive at
platforms and stop there, and that there is more than one platform. The communicative
intention is pre-processed in the conceptualizer, after which the contextual information
is passed on to the formulator in manageable chunks in the preverbal message. How this
information is passed on is not exactly clear. It is possible that we have some sort of mental
image of trains, platforms and arrivals which is then transformed into interpretable
information. (1992/2000:424).
De Bot goes on to adapt the original model. First, he gives two possibilities:
either there is a separate formulator and a separate lexicon for each language –
which makes it difficult to see how CS could take place – or there is one large
system storing all the information, “linguistically labelled in some way”, about
the different languages. The problem is then to understand how bilinguals keep
their languages separate. De Bot refers back to the early research which posed this
question in a very stark form (Kolers, 1963), and to more sophisticated recent
hypotheses about various types of “sub-sets” of linguistic elements described, for
example, by Paradis (1987). He comes down for an intermediate solution, i.e. that
the systems are partly unified and partly separate, depending on the closeness
between the languages and the speaker’s own proficiency in each. Links between
elements are strengthened through continued use, and in general, those between
elements of a single language would be stronger than between those in different
languages. In the case of regular CS, however, interlingual links could be just
as strong as intralingual ones. De Bot’s conclusion is that the lexicon and
phonological information are stored together, but that the formulator is languagespecific. Separate lemmas are necessary since these carry the syntactic information for correct formulation in each language. De Bot and Schreuder (1993)
suggest that “language” is one of the characteristics of a lexical element used in
the retrieval process: “In other words, in some settings words from the other
language can act as the next best solution to a word finding problem, much in the
same way that a near-synonym or a hypernym can serve as such” (p. 201).
De Bot agrees with Giesbers (1989) and also with Poulisse (1997) – whose
work concerns L2 learners – that CS as a “speech style” is made up of “performance switches” which are the “result of form characteristics being more or less
randomly linked to lemmas in the surface structure” (1992/2000:438). Such
switches contrast with those initiated at different levels in the production
process, for example those which are “triggered” by words which are similar
in form in both languages, or those which are due to unequal command of the
two languages, or lack of immediate availability of the required word in one of
them. An explanation is offered for cases where CS appears to be entirely “nondeliberate”, such as that of Dutch immigrants in the USA or Australia who, after
living for decades in an anglophone environment, have difficulty speaking Dutch
Psycholinguistic approaches
and keep slipping back into English, often at the point where they encounter
words which are similar or identical in both languages. De Bot and Schreuder
(1993) suggest that this is a case of Dutch being very weakly activated (through
disuse), whereas elements from the second language, English, are so highly
activated that they cannot be inhibited (see also Clyne, 1991a; 2003). De Bot
admits, however, that it is not always easy to draw clear lines between the
different types of CS, echoing the difficulties faced by sociolinguists in classifying switches according to their likely type/origin/motivation (see Chapter 3).
To sum up, De Bot, based on Levelt, claims that macro-planning (in the
conceptualizer) is not language-specific, that bilinguals have a single lexicon
and a single articulator, and that lemmas are linked to various form-characteristics
of words. As the production process advances, the preverbal message contains
information about the language in which parts of utterances are to be produced.
Finally the relevant language-specific formulator is activated. Kecskes and
Moyer (2002), in their “Ethno-conceptual model”, add that culturally specific
concepts are also already present at the conceptual level which activates the use of
lexical items. CS is therefore not simply a syntactic phenomenon, but the outcome
of individual choices which are conceptually motivated, as well as contextual
constraints. At a linguistic level, as a result, conceptually congruent but grammatically different elements may be interchangeable. They use this to explain a
number of examples showing English verbs used with the subcategorization rules
of their Spanish equivalents (2002:13).
Clyne’s analysis of the psycholinguistic models supports a model in which
material from both languages is stored jointly, but with elements from the same
language being more closely linked. He also believes that “Various contact
phenomena support inhibition of the less active language” (2003:242). De Bot
concludes by pointing out that “the empirical basis for an evaluation of
a bilingual production model is rather small at the moment” (1992/2000:442).
His use of the literature implies, however, that he is open to testing the model
against naturally occurring data. If this happens, it could provide a useful
underpinning, in psycholinguistic terms, of analyses based on data collected
for sociolinguistic purposes.
Myers-Scotton and Jake: testing theories on naturally
occurring data
Myers-Scotton and Jake (1995/2000; 2001) were mentioned in Chapter 5
because, unusually, their work represents an attempt to straddle grammatical
and psycholinguistic concerns. The 4-M Model and Abstract Model attempt to
follow through Jackendoff’s (1997) proposals on the role of the mental lexicon
in connecting a theory of grammar with language processing and production.
Along with very few other studies (e.g. Karousou-Fokas and Garman, 2001;
Broersma and de Bot, 2006), they use naturally occurring CS data to test their
In explaining their “4-M” Model, and in putting forward the CP as a unit of
analysis (see section 5.3.3), Myers-Scotton and Jake remark that the conceptual
level is not made out of syntactic units, but that instead “Intentions activate
language-specific semantic/pragmatic feature bundles”, which in turn interface
with language-specific lemmas (2001:87). The problems involved in verifying
such a chain of events have been alluded to above. What is new here is that
morphemes are said to fall into four psycholinguistically distinct types (content
morphemes and three different types of system morpheme), whose different
roles in language production explain various “asymmetries” in existing datasets – for example, the near-absence of full English noun phrases in a Spanish
frame, or the difficulty of inserting English verbs in an Arabic frame. It is
unclear how the system copes with intra-linguistic variations in the patterns.
Karousou-Fokas and Garman (2001) revisited these proposals in relation to
bilingual conversations recorded among twenty-four Greek–English bilinguals.
They accepted that there is a fundamental processing distinction between openand closed-class items, but chose another segmentation unit for discourse, the
“text unit”, different both from the utterance, which relies partly on the tone
unit, and from the CP, which is more grammatically defined, to analyse their
naturally occurring CS. Significantly – and this is an important innovation –
they compared the number of instances of CS in different grammatical positions
with the absolute number of these positions in the data. Their results showed
that the “text unit”, along with the onset of a new turn, were both powerful
switching positions in their data. The only boundaries to actually resist CS were
word-internal ones. This can be seen as part of an ongoing search for the surface
linguistic units which best correspond with code-switching (see Backus in
section 5.6.3). Planning, on the other hand, can take place in a language other
than the one currently being spoken or comprehended – “Both languages can be
equally operative during the planning process” (p. 65).
Grosjean (1982) was a pioneer among psycholinguists in considering bilinguality to be the norm and for being interested in it for its own sake. As such, he is
particularly attuned to methodological issues which vitiate psycholinguistic
research with bilinguals (see section 6.4.1). A major contribution lies in his
development of the concept of language “modes”, which links in with a number
of findings about bilingual processing. The basic proposal is that bilinguals are
able to function either in one of two (or more, in the case of trilinguals)
monolingual “modes”, or alternatively in a bilingual mode. These modes are
on a continuum, and where the individual is situated at any given moment
Psycholinguistic approaches
(base language)
Figure 6.4 Visual representation of the language mode continuum. The
bilingual’s positions on the continuum are represented by the discontinuous
vertical lines and the level of language activation by the degree of darkness of
the squares (black is active and white is inactive). (Nicol, 2001: Fig 1.1.)
depends on the state of activation of each language (see Figure 6.4). This
activation is a function of context, interlocutor, topic and other factors.
The implicit assumption is that, at any given time, one language is the “base”
language and thereby “governs processing”. Myers-Scotton cites this work as
the origin of the Matrix Language Frame model, and given the propagation of
this concept, it is important to consider on what it rests. “(1) the pre-eminence of
one language over the other during language processing by bilinguals; (2) the
distinction between the behavior of system and content morphemes in speech
errors; (3) other distinctions regarding these morphemes in patients with
Broca’s aphasia; and (4) the possibility of each language having its own direct
access to a common conceptual store during CS production” (Myers-Scotton,
1993b: 46–7). The base language is therefore a hypothesis designed to explain
a range of disparate findings from psycholinguistic experiments, such as the fact
that when English words were embedded in French sentences, French–English
bilinguals showed a delay in recognizing the English words: “Of interest is the
‘base language effect’ (i.e. the first reaction of subjects is to think that the
language of the word is the context (base) language, or what is called the ML
here”7 (Myers-Scotton 1993b: 47). But there are other possible alternative
explanations for this finding, such as that French texts with English words
inserted in them may have been an entirely new experience for the subjects and
caused a degree of old-fashioned confusion. Reading texts in two languages
has been found to take longer than reading texts in a single language – although
in some circumstances frequent switching can apparently neutralize the problem. Dussias (2001), also measuring reading comprehension time, found that
My italics.
frequent CS in a corpus was easier for subjects to process than infrequent
CS. There is more to this argument, and the degree of activation must be tied
in with the degree of proficiency; but as we have seen above, activation is a
multifaceted concept which has several origins, both external and internal to the
speaker, and does not operate in a blanket fashion.
Grosjean allows for the base language to change frequently when the speaker
is in the bilingual mode (how frequently is not discussed). This means, in terms
of Figure 6.4, that a fourth position is not allowable, viz. one where a bottom
square, for language B, is also black, i.e. fully activated. Grosjean does allow
for both languages to be equally activated in simultaneous interpreting (where
one language is activated receptively and the other for production), or in certain
experimental tasks, e.g. where subjects are told that the verbal stimuli they are
given belong to either language, but does not explain why it is exclusively in
such cases that equal activation is possible.
The concept of language mode and the question of one-at-a-time v. dual
activation are two sides of the same coin, and Grosjean (1997; 2000) reports on
a variety of experiments, carried out both by himself and by others, where the
notion of language mode is put to the test directly or indirectly. The concept is
found helpful in explaining a variety of experimental findings, of which only
two examples will be given here.
(1) In a laboratory-based study, Grosjean (1997) manipulated the language
mode of French–English bilinguals, who were told they were taking part
in a “telephone chain” designed to test the amount of information that could
be conveyed from one person to another. They had to retell French stories,
containing English code-switches, to three different interlocutors, who were
not actually present but who were described to the subjects by means
of biographical sketches. The first was described as having a strongly
francophone background, the second as having a more mixed anglophone/
francophone background and the third as having a balanced linguistic background, i.e. using both French and English regularly. He anticipated that
the subjects would be in an increasingly bilingual mode with each of these
three imaginary interlocutors respectively. As expected, it was found that
the number of code-switches into English varied significantly depending on
which interlocutor was being addressed. In fully bilingual mode they hesitated significantly less. Grosjean concluded that the concept of language
mode helps to predict the extent and type of CS.
What the experiment shows is that, having as far as humanly possible
excluded all other factors which could influence the outcome, and in the
absence of any other discernible influences, bilingual subjects adapt the
balance of their bilingual output in order to fit in with characteristics of their
interlocutor. What the experiment cannot tell us is what these subjects would
have done in real-life situations. Faced with a real, visible interlocutor,
Psycholinguistic approaches
whom one might or might not empathize with, and given a lack of external
pressure – unlike in the experiment – to speak a particular language, what
language choices would these speakers have made? We do not know
whether they code-switch in their daily lives, and we are unaware whether
the different modes are representative of variations in their natural speech
patterns. In any real situation, speech accommodation would be one of the
relevant factors, but we know this to be a very complex aspect of behaviour
(Giles and Smith, 1979). The results obtained can only tell us what happens
in the rarified context of the experiment, and the relationship of those results
with real-life performance remains uncertain.
(2) The second experiment concerns the phonology of CS, an under-investigated
area. According to Grosjean, various experiments have shown a “base
language effect” during the perception of CS, that is to say that “there is a
momentary dominance of base language units (phonemes, syllables, words)
at code-switch boundaries that can, in turn, slightly delay the perception of
units in the guest language” (Grosjean, 1997:236) – it might in fact be more
appropriate to describe such effects as “carry-over” effects rather than “base
language effects” so as to avoid prejudging the base language issue. Despite
the reported “momentary dominance of base language units” in perception,
this effect was not reproduced in the productive sphere. Grosjean and Miller
(1994) found no phonological “carry-over” from one language to another in a
code-switched condition when bilingual subjects had to pronounce various
English and French phonemes. For example, they pronounced the stop
consonants /p/t/k/ differently depending on which language the word was
in, just as they did in monolingual mode (though see MacSwan (2000) and
Botero, Bullock, Davis and Toribio (2004), who did find some phonetic
effects in the immediate environment of code-switches, as well as some
convergence in VOT8 values).
Finally, a caution about the language mode hypothesis has been issued
by Dijkstra and van Hell (2003) who tested it in relation to trilinguals. The
hypothesis assumes strong sensitivity to top-down context factors, which can
change in a gradual manner depending on the environment (Grosjean and
Soares, 1986). In an experiment where Dutch–English–French trilinguals had
to perform word-recognition and lexical decision tasks in a monolingual Dutch
context, their performance was significantly faster when the Dutch words
had cognates, even if these were in their L3, French. The language mode, and
the external activation on which it depends, seemed to have no effect on
Voice Onset Time is a measure of the way in which the vocal cords operate when different
phonemes are pronounced, and is characteristically different from language to language:“The
duration of the time interval by which the onset of periodic pulsing either precedes or follows
release” (Lisker and Abramson, 1964:37).
performance in this case; the authors therefore argue for the notion of activation
to be applied in a highly specified and testable way in the future.
The concept of “language mode” developed by Grosjean is a useful way
of describing the fact that many bilinguals, or at least those who have the
opportunity to speak their languages in separate contexts, can be in a more or
less bilingual “frame of mind” at a given moment. It is not surprising that asking
them to perform tasks requiring bilingual skills – especially ones which would
never be called upon in their normal lives – causes a certain brief delay or
surface dysfluency, though the Dijkstra and van Hell experiment described
shows that such an effect cannot always be detected. What is not altogether
clear is what kind of repercussions the hypothesized “base language effect” has
on bilingual performance and whether such effects go beyond the immediate
“mind-set” which affects minute-by-minute behaviour.
Simultaneous interpreting is unlikely to be a unique instance of both languages being equally activated, and the conclusions of Karousou-Fokas and
Garman (2001), described above, support the possibility of dual activation.
Finally an electrophysiological study of reactions to Spanish–English CS
showed that, under some circumstances at least, processing a code-switch
may entail less disruption than processing an unexpected item in the same
language (Moreno, Federmaier and Kutas, 2002).
In this chapter, we have reviewed some of the psycholinguistic work relevant
to CS, highlighting some problems with extrapolating from the experimental to
the real situation. Naturally occurring CS has been very little studied from a
psycholinguistic perspective, although at least some psycholinguistic measures
(e.g. VOT) can be applied to it. A recent psycholinguistic experiment put to the
test Clyne’s (1967) notion of “triggering”9 and found that CS was significantly
more frequent in the vicinity of a cognate (of which most were proper nouns)
(Broersma and de Bot, 2006), thereby confirming the validity of Clyne’s
analysis; unusually, the data used was a corpus of self-recorded, informal
conversations involving three Dutch–Moroccan Arabic bilinguals.
There is a lack of comparative psycholinguistic work which could help
account for the variations found within CS, though some models (de Bot,
Green) do attempt to account for CS which is due to competence problems
and failures to control the languages properly. So far, findings fail to throw
much light on the processes, whether externally or internally driven, which
underlie CS in conversational contexts, though Green’s model, which allows,
According to Clyne (1967), certain words, particularly cognates, are liable to “trigger” a switch
into the other language.
Psycholinguistic approaches
under some circumstances, for words to reach the surface regardless of language, is potentially compatible with “switching as an unmarked choice”
(Myers-Scotton, 1983; Myers-Scotton and Jake, 2001). Notions of inhibition
and activation contribute more to an understanding of apparently random, and
lexically driven, CS than to that which translates conversational purpose, and as
we have seen, the relationship between bottom-up (or internal) activation and
the top-down (or external) kind is still under investigation. Meuter (2005) has
pointed out that both activation and inhibition should be seen as more complex
phenomena than they have so far. As an example of that complexity, experimental evidence shows that, generally speaking, it is harder for bilinguals to
inhibit their L1 than their L2; however, perhaps due to the greater inhibition
effort required, switching back from a weaker L2 to the dominant L1 is more
demanding than vice-versa. Furthermore, calculations of the “cost” of switching must be related to specific tasks, as they are not the same, for example, in
naming and in comprehension tasks.
The possibility of “dual activation” is only considered in a minority of studies
(e.g. Karousou-Fokas and Garman, 2001). Yet single or competing activation
begs a number of questions in relation to fluent bilingual CS. Highly mixed
varieties described by sociolinguists such as Boeschoten (1998), Meeuwis and
Blommaert (1998) and Gafaranga (2005) could lead to a reconsideration of the
base language paradigm which is associated with single activation. As Tracy
(2000:28) put it, the coactivation of dual systems is a normal phenomenon,
visible in the interpretation of irony, metaphor and ambiguity: there may also be
“cognitive levels of abstraction where requirements of individual grammars can
be more or less playfully suspended, possibly not unlike what we, as linguists,
do all the time when we (voluntarily, to be sure) create ungrammatical examples” (p. 29).
Such coactivation is discernible when a fluent French–English bilingual
remarked “That’s a bit ridiculeux”, knowingly using a form which is neither
“French” nor “English” (English ridiculous, French ridicule) – though it
was given a French ending – and thereby implicitly mocking his own fake CS
(unpublished example). The choice of educated – and usually non-balanced –
bilinguals for psycholinguistic experiments tends to minimize such manifestations. Where production, unusually, is the focus of attention, most utterances
are assumed to be the product of a sequential single-system output, i.e. to be
fairly clearly assignable to one language or another. Compromise forms, “bare”
forms and “covalent” elements are rarely even mentioned in the psycholinguistic literature (Woolard, 1999). We should look forward to psycholinguistic
studies of CS where the concept of separate and discrete “languages” is not
taken as a given, and where such elements are central, rather than peripheral,
objects of study.
Acquiring code-switching: code-switching
in children (and L2 learners)
As was the case in earlier chapters, the preference for an interdisciplinary
approach makes it necessary to be highly selective within the various contributing research traditions. The considerable literature on both monolingual and
bilingual child language acquisition which is relevant to the issues discussed
here, cannot be summarized in toto and the reader is referred for further background information to Fletcher and MacWhinney (1995), Bialystok (1991) and
de Houwer (1995). As in the psycholinguistic studies, a number of issues
addressed in the literature on bilingual children have a bearing on CS, whether
or not children’s CS is the topic of the research as such.
One such example is that of the critical period hypothesis, on which there is
a large number of publications. This hypothesis states that there is a “critical
period”, generally thought to end at puberty, during which children can acquire
full native competence in their mother tongue. Children who suffer accidents,
illnesses or deprivations affecting their language development prior to this
critical stage are generally able to recover full linguistic functioning, whereas
those who suffer such traumas beyond this point may never be able to do so.
Similarly, second or third languages can usually be acquired with full nativelike competence by young children, but there is evidence that this ability also
declines with age, i.e. there is an – admittedly less crucial – critical period for
acquiring further languages as well (Hamers and Blanc, 2000; Johnson and
Newport, 1989). Issues to do with maturation of the brain and lateralization
form the backdrop to this discussion (see section 6.5).
If learning a second language is qualitatively different from a certain age
onwards, then it is possible that CS by early simultaneous bilinguals is also
qualitatively different from that of later second language learners.
Unfortunately, there is as yet limited research on CS in L2 learners (though
see Poulisse and Bongaerts, discussed below) and a lack of direct comparisons
with double L1 bilinguals. We do have evidence, however, that adults can
learn to code-switch late in life – several of the examples in Chapter 3 are
taken from adult migrants who “acquired” CS along with their acquisition of the
Acquiring code-switching: code-switching in children (and L2 learners) 143
L2 (see section 3.3.2). CS can also result from attrition of one of the speaker’s
languages, as with the elderly Dutch and German speakers in Australia
described by Clyne (see section Interestingly, it can also be used
extensively by speakers who are not very fluent in one of the languages, when
the indexical value of that language is an important part of their identity. Besnier
(2003) illustrates this phenomenon among Tongan “Leitī”, a category of males
who adopt numerous traditionally feminine characteristics and orient their lives
towards a “modernity” which, for them, is symbolized by the use of English.
Research on bilingual children is generally carried out with little reference to
studies of bilingual adults. Some mention is made of the psycholinguistic
models discussed in Chapter 6, and occasional references are found to studies
of adult CS in a sociolinguistic context, usually in order to point out that the
mixing encountered in bilingual children is a completely different phenomenon.
Once again, the insights to be derived from comparative studies are underestimated. From the point of view of an understanding of CS, studies of
bilingual children provide many important insights. Numerous studies have
been carried out, documenting children’s CS in a comprehensive way. Our
knowledge in this area is therefore rather solidly based. To give one brief
example, many bilingual children are reported as producing words containing
morphological mixtures and mixed two-word utterances between around one
and two-and-a-half, even when the languages are presented to them in separate
contexts. There is usually a steady decline in such mixtures thereafter (Hamers
and Blanc, 2000; van der Linden, 2000; Vihman, 1985; Tabouret-Keller, 1969).
Issues in child bilingualism
Characteristics of bilingual children
The subjects of child bilingualism studies are not generally typical of the
population at large. The majority are “elite bilinguals”, in many cases the
offspring of the linguistic researchers or their acquaintances, studied longitudinally over a period of months or years in the family context (Leopold,
1939–1949; Taeschner, 1983; Fantini, 1985; Saunders, 1982; Deuchar and
Quay, 2000). The advantage of this method is that the child’s speech is thereby
recorded in a natural setting, with access to someone who fully understands what
they mean. The disadvantage is that the children are likely to be at the top end of
the scale, both cognitively and socially (for a discussion of the particular features
of such bilinguals, see de Mejía, 2002). We lack studies of ordinary, workingclass children brought up in native plurilingual environments, who constitute the
majority (exceptions include Zentella, 1997; Tabouret-Keller, 1969).
Among these well-educated, middle-class bilinguals, there are a number of
variations: each parent speaking their (different) mother tongue to the child;
both parents speaking one language and the environmental language being
different; both parents speaking a language which is not their mother tongue
to the child, but speaking their mother tongue to one another, etc. (Romaine,
1995). Each of these situations leads to different degrees/types of separation of
the languages in the child’s input. This is important because one of the main
factors which we need to know about in order to understand the linguistic
behaviour of bilingual children is what type of input they receive from their
parents or carers. Most of the studies referred to below concern the development of early simultaneous bilinguality, but the variety of input – and
outcomes – is all the greater when one also takes account of consecutive
bilinguality. For a discussion of the differences, see Hamers and Blanc
(2000: 65–72).
Most of the bilingual children whose development is described in the
literature are presented by their parents with two languages which are deliberately kept separate, at least up to a point. The extreme version of this policy
is known as Grammont’s Law (1902), which advocates each language being
spoken on an exclusive basis by each of the children’s carers. More recently,
this principle has become known as OPOL – One Parent One Language
(Barron-Hauwert, 2004). Children whose parents decide to “manage” their
children’s bilinguality in this way are the majority of those described in the
research literature, but they are undoubtedly in a minority of bilinguals
worldwide. However, parents who claim to be strict followers of OPOL
often turn out, when their speech is systematically observed, to be subconsciously code-switching in spite of their best intentions (Lanza, 1997; JuanGarau and Pérez-Vidal, 2001; Goodz, 1989). The underlying basis of the
belief that OPOL is the best way to bring up bilingual children is that, in
many cases, it is the children’s ability to keep the languages separate which is
valued, rather than their ability to exploit the interaction of the two varieties.
As Lanza puts it, “Bilingual awareness is implicitly defined as formal separation of the two languages” (1992).
Most of the studies concern children whose languages are relatively closely
related (exceptions include Gawlitzek-Maiwald, 2000; Mishina, 1999; Vihman,
1985). This limits the opportunity to observe what happens in cases of structural
conflict between the languages. Ideally, we would want to carry out comparisons between children learning more or less distantly related language pairs in
order to test the effect of the linguistic factors proper on CS (the same is true of
adults – see Chapters 5 and 6).
Some problems with studying children’s code-switching
Most child language studies involve an assessment of the role of input from
adults, carers and the environment generally on the child’s learning and
Acquiring code-switching: code-switching in children (and L2 learners) 145
linguistic choices. In the case of bilinguals, and of CS in particular, the
question of what is learned and what is the result of more or less spontaneous
development is particularly crucial, so we will come back to this issue in
section 7.7 below.
Another, related challenge in the study of children’s CS is to sort out developmental from non-developmental issues. The mixtures produced by children
in a bilingual environment alter in type as they grow up, owing to their maturing
metalinguistic capacities, as well as their increasingly sophisticated linguistic
competence. Several studies (e.g. Meisel, 1989) state or imply that the early
manifestations of mixing found in such children are of a quite different nature to
“adult CS”, and that, whereas the latter is a matter of pragmatic competence, the
former is a function of the child’s developing language system and incomplete
differentiation between the two varieties. Some of the evidence we will discuss
does show children’s CS patterns changing over time in response to increasing
linguistic sophistication. But there are degrees of purposefulness in adult CS as
well as that of children. CS in adults can be related to uneven competence in the
two varieties, and is therefore not always a manifestation of pragmatic intent
(see section 1.1, example 2).
In fact, any assumption that adults can “distinguish two systems”,
whereas children cannot, needs to be examined critically. Some adults can
do this in formal or experimental tasks, or under particular circumstances of
production, but not every instance of CS entails that the speaker can make
such distinctions (see section Both sociolinguistic and psycholinguistic studies described in previous chapters suggest bilinguals’ ability –
or desire – to separate their languages depends on many factors, including
their education, the norms of the community where they live and the type of
It is doubtful whether there is, or could ever be, a reliable way, in adults or
in children, of distinguishing “deliberate” from non-deliberate CS except in a
small minority of cases (for example, those involving self-corrections or
certain types of flagging). At one end of the scale, some instances of CS
appear to be “slips of the tongue”, and at the other end of the scale, some
switches have a very clear conversational function, for example addressee
specification, or following the we code/they code pattern (see section 3.5.1.).
In between the two extremes lie the majority of CS utterances, where it is
unproductive to ask the question Why? and expect to find an answer based
on personal intentions. The “reasons” have to do with the patterns we are
exposed to, with conversational factors and with our psyche. At most, one can
say that there is a “scale of deliberateness” which can be applied to CS,
whatever the speaker’s age, but that with the exception of the extreme ends of
the scale, placing individual examples on points along it remains a matter of
(occasionally enlightened) conjecture.
Terminology again
By a certain age, bilingual children acquire the ability to code-switch for the full
range of functions used by adults, but, until that stage, the mixtures which they
produce are related to their linguistic development and to the degree of their
awareness that there are two varieties in their environment. It has therefore been
argued that it is more appropriate to use a different term, such as “mixing”, for
young children’s alternations (Meisel, 1989).
As with adult CS, studies of bilingual children often begin by explaining how
various terms will be used. While CS is, in this field too, a commonly used term,
several authors refer instead to code-mixing, or even just mixing (Boeschoten &
Verhoeven, 1987; Paradis, Nicoladis and Genesee, 2000), and as with adult CS,
the lines between its different manifestations are drawn in different ways.
Meisel (1989) defined “mixing” as the fusion of two grammatical systems,
and CS as a skill in the bilingual’s pragmatic competence (p. 36). Others (e.g.
Comeau, Genesee and Lapaquette, 2003) use “code-mixing” as the generic
term. Perhaps as a way of avoiding these endlessly repeated terminological
debates, the various contributors in Döpke (2000) refer mainly to “crosslinguistic structures”.
The effort to draw clear lines between different sub-categories of CS faces the
same difficulties in children as we discussed in relation to adults in Chapter 1.
On the whole, child language researchers have adopted a more pragmatic
approach than sociolinguists, using terms such as mixing, code-mixing or
code-switching fairly loosely, and specifying what phenomena they are focusing on in the particular study, e.g. lexical mixing. Many studies are concerned
with the earlier stages of acquisition and consequently with “mixing” at the oneor two-word stage, or even with phonological separation and mixing before that
(Deuchar and Quay, 2000). Some authors refer to “borrowing” in children, but
use the term in a different sense to the sociolinguistic studies, where it has a
diachronic dimension – although, as we have seen, the usage there is far from
consistent. In child language, “borrowing” tends to be used in relation to the
early stages, when children use words from either language because they lack
translation equivalents (Genesee, 1989). Code-switching for lack of alternative
also occurs in adult speech, whether the word is temporarily unavailable for
processing reasons or because it has never been acquired. Other terms which
occur include “interference” and “loan-blending”; the latter refers to using a
word from the lexicon of one language and adapting it morphologically to
the other language, and is considered perfectly systematic by some authors
(e.g. Hamers and Blanc, 2000:58), whereas for others, who subscribe to the
free morpheme constraint (see section, it represents an aberration.
Examples include such phrases as Ich hab ge-climbed up ‘I have climbed up’
Acquiring code-switching: code-switching in children (and L2 learners) 147
with the German regular participle prefix ge- added to the English participle
climbed, spoken by a two-year-old (Tracy, 2000); and I say we don’t go to trou-s
in France anymore,1 spoken by a twelve year old (unpublished example). Trou
was pronounced with French phonology but with the -s of the plural sounded as
in English.
Terminological mismatch, along with the fact that the child studies generally
focus on a significantly different type of production to the adult ones, should not
deter us from considering their relevance to issues addressed elsewhere in CS.
First, we have seen how one type of bilingual behaviour merges in with another
(section 2.2.4), and what is relevant to one type of bilinguality is often relevant
to other types as well. Second, as the old saying goes, “the child is father to the
man”. Adult code-switchers develop out of child bilinguals (as do adults who
keep their languages separate, of course), and although adult CS may result
from social conventions which an individual speaker encounters relatively late
in life (as in migration), the seeds of the ability to code-switch are planted in
childhood. There is therefore much to be said for trying to understand the
“ontogenesis of bilinguality”, to use Hamers and Blanc’s term (2000). For
example, if bilingual children, from the earliest stages, can separate their two
languages as and when necessary, it is easier to understand how, later on, the
ability to manipulate the two varieties in different circumstances might be a
subconscious skill.
One system or two – the wrong question?
For the terms “mixing” or “switching” to be meaningful, there must be two
identifiable, separate entities to mix. As with adult CS, we must bear in mind the
difference between the observer’s point of view and that of the speaker (see
section 1, example 1). The adult’s observation that a child’s utterance contains
elements of, say, Japanese and elements of French, may not correspond with any
dichotomy from the child’s point of view. There is no a priori reason why a
young bilingual child should be aware that the words, expressions and sentences which they hear around them are labelled as separate “languages” in the
outside world.2 This does not imply unawareness of the differences between
the varieties they hear. As we will see, this awareness has been detected from the
earliest age. But being aware of the differences – and able to reproduce them – is
not the same as having internalized rules as to what can be combined and what
A trou – literally a ‘hole’ – is a term used to mean an isolated place where nothing much happens.
The speaker was complaining about a long – and, to him, pointless – car journey through France.
This is separate from the issue of when children acquire the adult names for languages. Some
children use names like “Mummy’s language” and “Daddy’s language” before they acquire
language names, but the point here refers to the early stages before any such labelling is found.
should be kept separate. This more metalinguistic ability, which includes the
learning of norms appropriate in the environement, is likely to proceed at a
different pace (Tabouret-Keller, 1969).
What does “mixing” mean in children?
Children from the youngest age can discriminate between patterns. They have
the ability to pick up co-occurrences, e.g. of French phonemes with French
words, and will tend to reproduce those co-occurrences. If this were not the
case, the “mixing” which has been identified in young bilinguals would be more
extreme than it is usually found to be. One might wish to make a distinction
between mixing and simple confusion. The latter might depend on many
factors, notably the particular language combination which is involved. For
example Paradis (2000) suggests that the relatively variable word order in
Dutch and German explains why some studies using these pairs have found
transfers due to confusion between the two, and Schelletter (2000) shows how
the contact situation motivated a German–English bilingual child to analyse
negation in both languages along the German – more adverbial – model. Meisel
(1989) refers to the fact that children sometimes use translation equivalents as if
they were synonyms as a symptom of “fusion”.
Certainly at the one- and the two-word stage, i.e., in practice, well into the
child’s third year, it is not possible to talk in terms of a “base language”, as
many two word utterances consist of one word from each language, such as où
car? ‘where car?’ from a French–English bilingual two year old (Genesee,
2001:155). Even when the child’s sentences begin to include more morphemes, the “language of the conversation” criterion and the morpheme
count criterion are often in conflict: “There may well be utterances without a
‘definite language’ but rich in multiple representations instead” (Tracy,
An extended example from this study, along with a selection from Tracy’s
commentary, illustrates several typical aspects of mixing in young bilinguals.
The child, Hannah (two years, three months), a German–English bilingual, is
trying to get her mother to help her strap her doll into her buggy. This example is
reproduced as in the original text deliberately without signalling the language
changes with a different font etc., as the distinction between the languages
forms the subject of the commentary; therefore only a gloss has been added
(\ signals falling intonation).
Example 1
(a) die dolly [ənstræpən] \
the dolly strap-in INF
(b) die dolly [əntræp] \
the dolly ..trap
Acquiring code-switching: code-switching in children (and L2 learners) 149
(c) dəs eins-[stra:p i:n] die puppe \
it in- strap in the dolly
(d) die ein-[stra:p i:n] die dolly \
it in strap in the dolly
(e) die mama helf mir [tap] it [i:n] \
(the) mummy help me strap it in
(f) mama (=voc.)
[thap] it [i:n]\ die dolly \
mummy [=voc.] strap it in \ the dolly
The child starts with what looks like a German VP, (a)–(b), with the verb in the latter
case already on its morphological way to an English infinitive, the form required for the
English constructions with which she ends up in (e) and (f). In (c) and (d) we witness
some blending of both German and English with respect to syntax and morphology.
Throughout, the long vowel of in appears to retain phonological properties of the
separable German particle ein (including stress).
Note that strap is not a verbal stem in German, but this is an accidental, not a
systematic gap. Also, in this particular instance, there is no other equivalent German
expression with the separable prefix ein. The most idiomatic translation for the set (a) to
(f) into German would require anschnallen (or festschnallen), and actually, for all we
know, the an might well be on its way in the “grey”, ambiguous onset of the first syllable
of the verb in (a) and (b). On top of the contrast involving the positioning of the verb with
respect to its complement (OV vs. VO) and the contrasting morphological design of an
English v. a German infinitive (stem +Ø v. stem + -en) there is, then, from the child’s
perspective, an additional conflict involving a (virtual) German particle verb (einstrappen) and an English phrasal verb (strap in). The overall tentative pattern shows two
things: uncertainty about the home base of the verb and, nevertheless, expertise in
conjuring up the contexts appropriate to both languages, with the blends in (c) and (d)
offering us a particularly good glance at online competition and conflict resolution. This
resolution is not a permanent option for the child’s grammars, which can contain only
VO or OV. It is impressive how early bilingual children become experts at handling what
they perceive to be crucial design features of each language. (Tracy, 2000: 30–31)
The interesting thing about this example is that although the child is mixing
her languages at different levels, as Tracy points out she is clearly aware of
various differences between the two languages. For example, she has a translation equivalent for the dolly ‘die Puppe’ and she is aware of English wordorder rules (line (e) as opposed to the rest). Even if we describe her as having
“two systems” here – which seems a pointless oversimplification – she seems
more concerned with how to combine them, so as to maximize her resources,
than how to separate them. As Meisel (2000) remarks, the question of whether
the child is applying one system or two cannot be separated from the question of
when grammatical processing starts – i.e. when it takes over from the semantic/
pragmatic principles which, it is generally assumed, serve to organize early
child language.
More recent studies have tended to incorporate the “hypothesis of independent development” (Genesee, 1989) and to address detailed questions regarding
the type of separation or interaction between the languages. For example, Yip
and Matthews (2000) use both quantitative and qualitative methods to examine
syntactic transfer effects from Cantonese to English in their bilingual son
between the ages of one and six. They support the idea that two systems are
developing independently, but do note a number of transfer effects from the
dominant language, Cantonese, to English. By contrast, Juan-Garau and PérezVidal (2001), observing a Catalan–English boy in similar circumstances (OPOL
was practised in both cases), found that the aspects of word order which they
were observing (notably subject realization) developed quite separately with no
transfer being evident. More comparative research is needed to establish under
what linguistic and other conditions transfer phenomena are present or absent in
bilingual children’s speech.
Discrimination in receptive and productive abilities Receptive Various experiments show that babies are aware of differences between languages from the earliest age. Thanks to a technique which
conditions the infants to turn their head in response to a change in continuous
auditory stimulation, it can be shown when they are differentiating between
different languages spoken in their environment (Bosch and Sebastian-Gallés,
2001). Various studies have thus been able to show that babies from a few days
to a few months old can discriminate the sounds of their mother tongue from
those of another language (Hamers and Blanc, 2000). Children brought up in a
bilingual environment are, it would seem, better at discriminating between their
own language and another, totally unfamiliar, language than children brought
up monolingually (Eilers, Gavin and Oller, 1982). The ability to discriminate is
based on the characteristics of the signal itself – whether or not the parents are
practising OPOL is secondary. Productive As regards production, even before children produce
their first word, their babbling differs depending on the language of their
interlocutor (Giancarli, 1999). Maneva and Genesee (2002) found that 10–14month-old bilinguals babbled in different phonological patterns with their
French-speaking father and English-speaking mother. This being so, it is not
surprising to find that when they start speaking, bilingual children adapt to the
language of the interlocutor. As they progress, there is increasing separation of
the varieties according to the interlocutor. Growing awareness is demonstrated
by the fact that children start asking for translations, which seems to occur
around the age of two (Hoffmann, 1991; Vihman, 1985). However, considerable inter-individual differences have been noted in the level of bilingual
awareness and therefore in the amount of mixing which takes place (Arnberg
and Arnberg, 1992; Genesee, Nicoladis and Paradis, 1995). For example,
Acquiring code-switching: code-switching in children (and L2 learners) 151
Morita (2003) describes the development of two Japanese siblings living in the
USA. One avoided using Japanese terms of address, which are far more complex than the English system, until he was able to deduce the right forms from
his mother’s input. The other got round the problem by using English you in
Japanese utterances.
Recent studies often take as their starting point the contention in Volterra and
Taeschner (1978) that, in the earliest stages, children exposed to two languages
have a “single system”, which they then gradually learn to differentiate. Volterra
and Taeschner identified three stages: first, a single lexical system including
words from both languages; second, two lexical systems but a single set of
syntactic rules; and finally, differentiated lexical and syntactic systems. This
“single system” hypothesis was supported in various forms by other authors
around the same time, e.g. Vihman (1985) and Redlinger and Park (1980). An
interesting parallel may be drawn with the psycholinguistic explanations of
CS in terms of a single-switch system, which were then replaced by a dual,
and finally by a more complex and differentiated set of procedures (see
section 6.5.2).
Over the last twenty or so years, the “single system” hypothesis has been
criticized on several different counts. It is only possible to give a selection of
these discussions here, and we will concentrate on three major case-studies,
namely De Houwer (1990), Lanza (1997) and Quay and Deuchar (2000). As De
Houwer points out, it is far from clear what type of evidence would be adequate
to confirm either a single-system or a dual-system hypothesis (1990: 49). Such a
discussion is subject to the same ambiguities as those which surround the use of
the term “language” (see Chapter 1).
(1) De Houwer’s subject Kate, a Dutch–American English bilingual, was
observed from 2;8 (two years, eight months) to 4;0. Although her CS
was found to be skilled and smooth, in the earlier stages of the observation,
Kate did occasionally address monolinguals in the “wrong” language or
code-switch inappropriately. Interestingly, her self-repairs increased dramatically in both languages at the same time – including the “repairing” of
inappropriate switching, at the age of 3;1 to 3;8 (1990:113). While De
Houwer considers that there is a maturational element involved, as evidenced by the relatively sudden onset of the self-repairs, she also stresses
that plenty of evidence was found for separate morphosyntactic development, similar to that in monolingual children, in both of Kate’s languages.
The latter finding confirms results from a group study by Kessler (1971),
who remarked that: “Structures common to two languages are acquired in
approximately the same order and at the same rate by bilingual children …
The sequencing of specific structures is dependent on the extent of linguistic complexity, describable in terms of case grammar relations … The most
linguistically complex structures are those acquired last” (pp. 93–94).
Döpke (2000), however, points out that just because the vast majority of
structures produced by bilingual children are acquired in exactly the same
way as by monolingual children, this is not a reason to ignore the crosslinguistic structures, as they can still be systematic, as opposed to “performance errors” (see section 1.8).
(2) Lanza (1997) studied two two-year-olds simultaneously acquiring
English and Norwegian. Various differences appeared between the
acquisition patterns of the two children, one being only slightly dominant in Norwegian, the other more markedly so. One family practised
OPOL, the other used more CS. As far as the formal aspects of the
children’s mixing were concerned, Lanza found that, in the early stages,
the children’s dominance in Norwegian expressed itself in the fact that
they used a considerable number of Norwegian function words with
English lexemes, but not vice-versa. Bolonyai (1998) found similar
linguistic evidence of the dominance of one language over the other
in a Hungarian child who moved to the USA. As English became more
dominant, she claimed that a “turnover in the Matrix Language” had
occurred (see section 5.7).
Dominance in one language does not, however, always lead to such a
pattern. An English–French bilingual three-year-old used lexical importations from his dominant language, English, in sentences whose grammar
was entirely French, as in: Ne me scrabbe pas comme ça – où est le sponge
de Nicolas? ‘Don’t scrub me like that – where is Nicholas’s sponge?’ where
‘scrub’ and ‘sponge’ were pronounced as French words (unpublished example).
Using the English words presumably seemed a better solution than not
expressing his desired meaning at all. Many adults also get away with doing
this and are probably well enough understood, thanks to the redundant
quality of language. A parallel can be drawn here between individual
bilinguality and group/community bilingualism. Thomason (2001) asserted
that, on the whole, it is the minority group which becomes bilingual and
adopts features from the dominant group (see section 2.2.1), but that the
majority sometimes also contribute to language change in the minority
language. This child’s use of his stronger language (English) in the context
of his weaker one (French) can be likened to this process at an individual
level. Multiple factors are always at play and the pattern of mixing depends,
among other factors, on exactly where the individual bilingual’s lacunae lie.
Overall, Lanza found the most significant factor in the separation or
mixing of the two languages to be the type of bilingual context which each
of the parents encouraged. Thus the little girl, Siri, did more mixing with her
father when speaking Norwegian than with her mother when speaking
English, even though she was dominant in Norwegian, because Siri’s
mother’s strategies discouraged mixing. The fact that the children in her
Acquiring code-switching: code-switching in children (and L2 learners) 153
study were able to differentiate the two languages did not preclude mixing
in certain situations. These results are confirmed by numerous reports
(Genesee et al., 1995; Mishina, 1999; Shin and Milroy, 2000).
(3) Crucial evidence about children’s ability to discriminate is provided in
various studies of infant and child phonology. Deuchar and Quay (2000)
investigated Voice Onset Times3 and other phonological differentiators in
an infant acquiring English and Spanish, and found an ability to differentiate from the earliest stages. Further examples include:
Brulard and Carr (2003), who report that a French–English bilingual
child used strategies identical to those of monolingual children both in
French and English words from the earliest stages of his producing
words (reduplication in French and consonant harmony in English);
Gut (2000), who detected no cross-linguistic influence in the acquisition
of question intonation of a German–English bilingual child between 2;5
and 4;3.
Johnson and Lancaster (1998), whose study of a Norwegian–English
bilingual boy found that, although his phonemes differed from those of
monolingual children in both languages, overall he clearly distinguished
sound production in each one.
As Deuchar and Quay (2000) are at pains to point out, a bilingual child is only
in a position to respond differently in the two languages once translation
equivalents have been acquired. Before that stage, it is pointless to think of
there being any “system” as such, even less two competing ones. As in the case
of the children observed by De Houwer and Lanza, their study showed that the
language of two-word utterances was matched to the linguistic context – with a
few exceptions in the form of words which were used indiscriminately, despite
context (e.g. more and its Spanish equivalent, más). From the point of view of
syntax, in a longitudinal study of two German–French bilingual children,
Meisel (2000) found that they applied different word order in the two languages
as soon as they started to produce multi-word utterances.
So rather than focusing on the question “Is there one system or two?”, as
Deuchar and Quay suggest we should concentrate on how and when differentiation occurs in the child’s language. This varies according to the aspect
being considered (lexis, morphosyntax or phonology), and probably, as we have
seen with Lanza, in accordance with the dominance of one language over
another. There is no point in describing children’s utterances as mixed when
the child has no choice because they have not yet acquired the translation
See Chapter 6, footnote 8.
The case-studies therefore concur on a number of crucial points. There is
considerable evidence that even very young children are able to differentiate
between languages. Allowing for individual variations, which occur in monolingual children as well, the bilingual children studied learned each of their
languages in the same way and at the same rate as their monolingual counterparts. Finally the tendency to mix or not to mix, after the very early stages, is
partly individually determined, regardless of similar input.
Do children obey constraints?
In section 5.3.1, we saw that the constraints on CS which have been identified
are not categorical, but merely reflect tendencies which develop in particular
circumstances. If there are no absolute grammatical constraints on adult CS,
then this is probably the wrong question to ask for children also. However
several authors have addressed this question in relation to children, and as in the
case of adults, the conclusion depends on the particular data-sets which are
examined. Genesee (2001) concluded from a survey that children’s CS obeys
the same constraints as that of adults. Vihman (1985; 1998) found that most of
the constraints implied by Myers-Scotton’s MLF model are observed in her
corpus, though she remarks on a few inclusions of system morphemes from the
embedded language with grammatical agreement in Matrix Language. On the
other hand, Bader and Minnis (2000) give examples from an Arabic–English
child violating Poplack’s constraints. Tracy (2000:19–20) gives examples of
many instances of child language where a base language cannot be assigned, for
reasons outlined in section 7.4.1 above.
Few of these studies directly compare children’s speech with that of the adults
in their immediate community (though see Mishina, 1999). Such information
would seem highly relevant to understanding the nature of such constraints, if
they exist, and to whether they are universal/inborn or the result of input. More
studies involving both adult and child CS would therefore be welcome in the
The role of input
More specifically, the role of linguistic input in children’s acquisition is
crucial. The Chomskyan position that language is an innate human faculty
finds support from what is known as the “poverty of stimulus” argument,
meaning that the language heard by children in their environment is generally
a far from perfect model, yet all normal children nevertheless manage to
acquire full grammatical competence in their mother tongue by the time
they are a few years old. De Houwer (2009) points out that the study of
bilingual children confronts us particularly starkly with the question of
Acquiring code-switching: code-switching in children (and L2 learners) 155
what role input plays in child language acquisition. For example, how, and to
what extent, is non-balanced input in each of the varieties reflected in the
child’s own competence?
“Elite bilinguals”, as we have seen, are often actively discouraged from
mixing their languages by their carers. However, we know that, in a monolingual context, explicit adult corrections of the child’s language are more or
less ineffective in the early years (De Villiers and De Villiers, 1979:108–111).
Furthermore, the injunction to use only one language at a time could only have
an effect on children to the extent that they were already consciously aware of
speaking two different languages. The gaining of such awareness is a gradual
process, as Taeschner (1983) illustrates: whenever her German–Italian bilingual
daughters addressed her in Italian, she would reply: Wie bitte? ‘Pardon?’ in
German, in order to get them to repeat what they had said in German. This
policy was effective at first, but broke down when the girls realized that their
mother understood Italian perfectly well.
Various studies, notably Lanza (1997), emphasize that each parent conveys,
through their discourse practices, what proportion of each language is expected,
and whether CS is acceptable; the extreme version of this being the OPOL
policy. Deuchar and Quay (2000) relate in statistical terms how much of each
language the child in their study is exposed to. But they rightly point out that
language separation by the parents is not necessarily the sine qua non for the
children to learn separation. Eventually, children pick up norms of the society
where they live; the question is how/when they become sensitive to these norms
and able to reproduce them.
Comeau, Genesee and Lapaquette (2003) claimed that many studies were
inconsistent or inconclusive regarding the effect of parental mixing on children.
They therefore tested six bilingual children (average age 2;4) in a carefully
designed experiment where bilingual assistants deliberately manipulated their
own mixing rate. In almost all cases, the children adjusted to match their
interlocutors on a turn-by-turn basis. They concluded that an earlier study of
parent–child interaction, which showed no correlation of this type, was flawed –
observing a child with their parent, as in the earlier study, may not be ideal
because their behaviour may be affected by socialization patterns within the
family which are not obvious to an observer. Goodz (1989; 1994), who carried
out another group study, longitudinal this time, of seventeen French–English
bilingual children, also found the children’s mixing to be related to mixing by
parents. The ability of even very young children (2;4 to 2;7) to adjust their rate
of CS to that of their interlocutor was again tested experimentally and found to
be significant, even when this meant adjusting their speech to use a higher
proportion of their weaker language (Comeau, Genesee and Lapaquette, 2003).
Genesee (1989) aptly points out that such results weaken the argument that
children’s mixing reflects an undifferentiated language system.
One important aspect of input is which particular language combination the
child is presented with. Paradis (2000) has suggested that the various studies
focusing on Dutch–German acquisition may have highlighted the effects
of cross-linguistic interaction, as both languages have a relatively variable
word–order and are, of course, closely related, thereby potentially causing
confusion. Conversely, Sinka (2000) found that Latvian–English bilingual
children mixed mainly at the lexical level, and suggested that the very
different typological characteristics of the two languages enable the child to
separate the two language systems at an early stage and so to produce
language-specific structures. Vihman (1985) claims that certain aspects of
children’s switching may be related to qualities of the languages themselves.
She ascribes her son’s use of English function words to the fact that they were
simpler and more salient for him than Estonian ones (see the notion of
“atttactiveness” in language change, section 2.4). Juan-Garau and PérezVidal (2001) stress the need to look at the interplay of parental input and
the child’s own pragmatic/linguistic development in order to understand how
the child’s CS decreases with age – though Meisel (1989) had talked of a
U-shaped curve: lots of mixing to start with due to limited competence in the
two languages, then fewer mixes as competence increases, then finally the
acquisition of adult-like CS abilities.
One of the few studies of a working-class child being socialized in a family
and a society where CS is the unselfconscious norm is Tabouret-Keller
(1969). By the age of six, her subject, Line, was code-switching between
French and Alsatian in a manner quite in tune with that of the adults who
surrounded her. This is doubtless representative of the majority of bilingual
children worldwide, whose repertoire is shaped by a combination of parental
input and, increasingly as they grow up, by all the parameters of their own
community’s plurilingual behaviour. As Zentella (1997) points out, subgroups within the community also create their own language and their own
repertoire and do not merely reproduce those of the preceding generation:
“When Paca, Lolita, Isabel, Blanca and Elli speak, they reflect their participation in particular ethnic, racial and class networks via their standard and
non-standard Puerto Rican Spanish, AAVE, and lower working class and
standard New York English, and those dialects tend to influence each other”
(p. 271). Just as Zentella remarks in this context that a purely quantitative
sociolinguistic approach risks missing the individual and group meaning of
Spanish–English switches, so the rather “asocial” approach adopted in many
studies of child bilinguality is likely to underestimate the active creation of
meaning by children with the basic elements provided by the bilingual models
around them.
Acquiring code-switching: code-switching in children (and L2 learners) 157
Pragmatic uses of code-switching in children
Lanvers (2001) found socially determined CS at a very early age, as well as CS
for discourse functions such as emphasis or appeal to the adult. The most
frequent CS was classed as “developmental” and was due to lexical gaps. A
number of researchers have found CS to be a function of addressee from an
early age as well as developing further functions over time (Jisa, 2000; ErvinTripp and Reyes, 2005).
Halmari and Smith (1994), in a study of two Finnish–English bilingual sisters
aged 9;2 and 8;1 living in the USA, found that 74 percent of their switches
occurred when “negotiation talk” started. They noted that particular features
co-occur with negotiation, even when English was used for this function instead:
voice quality changes, interrogatives, imperatives, etc. – showing CS to be a part
of register shift. Shin and Milroy (2000) showed how Korean schoolchildren
in New York not only adapted their language choices to the preferences of their
interlocutor, but also used CS to negotiate the structural organization of the
Cheng (2003) carried out a group study involving sixty Malaysian preschool
children, twenty-eight Chinese and thirty-two Malay. Code-switching in this
group was linked to a variety of discourse strategies. Cromdal (2001), who
observed children at an English school in Sweden, found that CS was mainly
used as a way to resolve overlaps in the discussion, a change of language
enhancing the second speaker’s chance to take the floor.
Cognitive effects of code-switching
There is a large literature on the cognitive effects of bilingualism (see for
example Cummins, 2003), but, apparently, little work on the relationship of
CS to cognitive development. Vygotsky (1934) believed that language and
thought started developing separately, and that language contributed only
subsequently to the development of cognitive structures. Early bilingual
experience – and, possibly, CS a fortiori – could in theory thereby encourage
greater cognitive flexibility, through the greater use of the self-regulatory
functions of language (Peal and Lambert, 1962:14). Diaz (1983) pointed
out, however, that this involves various assumptions: (1) that bilingual children are thinking verbally in performing non-verbal tasks; (2) that they switch
from one language to another while performing these tasks; and (3) that the
habit of switching improves performance. None of these have been demonstrated yet.
Döpke suggests that the tension between the two systems which bilingual
children have to process has a positive role in their language acquisition. She
derives this from the fact that the bilinguals whom she studied chose the
structures common to their two languages (English and German) with greater
frequency than monolingual children (2000:94), which shows that they (a)
noticed and (b) were able to exploit the contrast between the two (see also
Hulk and Müller 2000:229). Similar observations, as we have seen in
Chapters 3 and 5, have been made with regard to adults. While this is not the
product of CS as such, but rather of bilinguality, it seems likely that an environment where CS is common would reinforce the sense of contrast and encourage
this process further. Sinka (2000), mentioned above, who studied children’s CS
between two typologically very different languages, Latvian and English,
considered that the different nature of the two languages enabled the children
to separate the two systems at an early stage and to produce language-specific
structures. Accordingly, the mixing observed was mainly lexical. Various
studies have shown that bilingual children have an enhanced ability, compared
with monolinguals, to deal with abstract concepts, which, it is thought, may be a
by-product of the early necessity to separate the signifier and the signified
(Hamers and Blanc, 2000).
Group studies
Many of the important studies in this field are individual case-studies, as we
have seen above, but there are also a number of group studies, often longitudinal, which reinforce the results obtained. Together, they give an opportunity – as
yet insufficiently exploited – to study the range of individual differences in
various aspects of bilingual acquisition.
Lindholm and Padilla (1978) studied eighteen Spanish–English children
aged 2;0–6;0, and found that the switches observed served both sociolinguistic and communicative strategies. Redlinger and Park (1980) observed four
bilingual children (two German–French, one German–Spanish, one German–
English) aged 2;0 to 2;8 and found that with increasing MLU4 values, the
percentage of mixed utterances decreased. Köppe and Meisel (1995) studied
five French–German bilinguals aged approx 1;0–4;0, and Meisel also drew
for several of his publications on the results of the DUFDE Project, which
involved observing children with one French and one German parent. Kessler
(1971) studied twelve Italian families with children aged 6;0–8;0, living in
Philadelphia. She found that, while bilingual children’s language development
MLU (Mean Length of Utterance) is a measure of the average number of morphemes in children’s
utterances, which serves as an indicator of linguistic development.
Acquiring code-switching: code-switching in children (and L2 learners) 159
closely followed that of monolinguals, they tended to acquire structures common to the two languages more readily (see Döpke above).
Code-switching in the classroom
While many children worldwide are exposed to two or more languages in the
home, many are also confronted with a second language at school, or take part in
some form of bi- or trilingual education, through design or otherwise. De Mejía
(2002) has described elite bilingual education, where, in spite of the prevalence of
purist attitudes in many contexts, much CS nevertheless occurs. De Mejía (2002)
and Lin (2001) give extensive examples of the use of CS in bilingual school
situations, the first in Colombia in private English–Spanish immersion schools
where CS is accepted, and the second in Anglo–Chinese secondary schools in
Hong Kong where it is officially frowned on, but is nevertheless flourishing. In
multilingual Singapore, CS is also an important tool for children in the learning
process. Both positive and negative views of CS in education have been expressed.
Martin-Jones has carried out extensive research on classroom CS (1995; 2000)
and has demonstrated what a widespread phenomenon this is, and the variety of
purposes it can serve. It may for example reflect language practices outside the
classroom; serve as an inclusive strategy where pupils are of varying language
competences; serve to encourage pupils’ acquisition of an L2 by ensuring that
they understand at least part of what is said without difficulty; and so forth.
Martin (2003) illustrates classroom CS in Brunei between Malay and
English, but also between different varieties of Malay – Brunei Malay,
Colloquial Brunei Malay and Bahasa Melayu – especially among the young.
Such switching involves moving along a continuum rather than categorical
switches, and this is reflected in classroom discourse between teachers and
students like that in example 2. This owes its existence equally to pedagogic and
discourse-structuring reasons:
Example 2
Now carbohydrate …kalu kamu makan
if you eat
mandangar tu tadi cakap teacher
listen to what I’ve just been saying
if you eat too much of these food. Apu maksud teacher sana..apu maksud
what’s my meaning here?
kalau makan terlalu banyak
if you eat too much
jadi lampuh
get fat
apanya kata teacher tadi
what did I say just now?
mambari lampuh
get fat
it is not good for your
for your ..health..for your health. not good for your health..you will become very
(Martin, 2003:78)
Similarly, Cleghorn (1992) shows how teachers in primary level science classes
in Kenya convey ideas more effectively when they do not adhere to an Englishonly instruction policy. Merritt, Cleghorn, Abagi and Bunyi, (1992) list four major
factors which account for CS in the classroom: (1) official school policy; (2) cognitive concerns; (3) classroom management concerns; and (4) values and attitudes
about the appropriate use of the languages in society at large. For an African
perspective see the papers in Rubagumya (1994). Further studies include Hurtado
(2003), Foley (1998), Camilleri (1996), Garrett, Griffiths, James and Scholfield,
(1994); and Garrett, Griffiths, James and Scholfield, (1994).
A glimpse of classroom CS in action is provided by McCormick’s notes,
taken while observing a class in a school in South Africa in an area known as
District Six:
Two children were quarrelling in Afrikaans about their fathers’ work. (The fathers are fruit
and vegetable vendors.) A child intervened in English and the teacher redirected the
comment to the group as a whole, using English. Then, in both languages, she invited
responses from the other children. She asked, among other things, whether they knew a
traditional vendors’ rhyme which was in Afrikaans. One of the quarrelling children was
encouraged to sing it. An interlude of physical exercise followed. It was directed in
English. Then a child started telling a story in English about how he had been bitten by
a spider. He used the word “bleed” instead of “blood”: and then the bleed came out. “The
teacher supplied the orthodox phrase but did so without drawing attention to the correction:
“Oh! So the blood came out!” Several children started talking at once about spiders, some
in English and some in Afrikaans. The teacher picked up some threads in Afrikaans but the
next child to hold the floor did so in English. He was shy and she encouraged him in
English. She was then distracted by an upheaval in one corner of the classroom and asked
one of the children in Afrikaans what was happening. (McCormick, 2002:141)
While the informal framework which prevails with this young class may be
particularly conducive to relaxed CS, the latter clearly serves a number of
important functions in differentiating between types of discourse and in allowing the teacher to fulfil different roles, from directing attention to including
shyer members of the class.
Code-switching in second language learners
CS has not been very extensively investigated in relation to L2 learners,
though see Poulisse (1997), Dewaele (2001) and Poulisse and Bongaerts
Acquiring code-switching: code-switching in children (and L2 learners) 161
(1994). Poulisse and Bongaerts (1994) tie in learners’ CS with language
production issues as described in Chapter 6. They make the point that for a
beginner L2 speaker, the L1 has a much higher “resting activation” than the
corresponding word in the L2 (see section 6.6.1). They looked at “unintentional” (i.e. non-flagged) CS in Dutch schoolchildren learning English, and
found that this consisted principally in the insertion of L1 function words in
L2 (English). Poulisse and Bongaerts claim this to be a result of processing
considerations and in particular claim it provides support for the “spreading
activation model” (Dell, 1986). In an interesting reversal of the sociolinguistic and grammatical findings, they point out that content words, being
chosen with greater care, are less likely to be switched. However it seems
unlikely that this pattern would be repeated in such a way as to confirm that
this was a universal for second-language learners. Other cases of codeswitching by speakers with very uneven competence – while they might
not be in the same specific situation as these learners – have produced quite
different results.
Lüdi (2003) also discusses whether L2 learners code-switch. He points out
that it is a well-known communicative strategy for non-native speakers to use
their L1 (or another language) to get round communicative stumbling blocks,
a phenomenon which he calls translinguistic wording, but that balanced
bilinguals use this strategy also, as even they are bound to have some lexical
gaps. Such phenomena are generally flagged by dysfluencies and hesitations.
At the same time, learners may “code-switch” in the same way as balanced
bilinguals when they alternate fluently between their L1 and L2, in particular
when they are speaking their L2 and need lexical elements from their L1, as in
example 3, from a native speaker of German/non-native speaker (NNS) of
Example 3
nns: et il y a un petit moteur oui oui qui tire le cocon et il y a la vorrichtung
and there is a little motor yes yes that pulls the cocoon and there is the device
il y a de grands mast de stahl
there are tall masts of steel
(Lüdi, 2003:178)
Lüdi’s point that there is a continuum between CS and translinguistic wording is well taken; however, his suggestion that CS can be meaningfully
explained with reference to the notion of Matrix Language is subject to the
problems which have been discussed in Chapters 5 and 6. Finally, in a study of
Igbo-Nigerians learning Italian, Goglia (2006) points out that learners whose
native varieties involve unmarked CS may well transfer the switching habit to
the language they are learning, thereby increasing their resources in that
Conclusion: code-switchers or code-breakers?
The conclusions we can draw from studies of bilingual language acquisition are
complicated by the fact that children can be presented with two (or more)
linguistic varieties in so many different configurations. Different methods of
observation, appropriate for different ages and circumstances, are also partly
responsible for the variety of results obtained. These factors mean that comparability between the studies is an issue. It is not surprising that widely different results are found with respect to many of the questions asked, such as
whether children transfer elements from their dominant language to their
weaker language only or whether the process is two-way. Also, as with adults,
there appear to be no studies yet of CS in relation to personality and cognitive
style, although there are strong indications that these affect the quantity and type
of CS. Children also vary in the extent to which they pick up, or care about,
linguistic purism on behalf of their carers.
The question which runs through this book, i.e. what “a language” is, and
how CS can help us to answer it, is powerfully raised by studies of bilingual
children. Many researchers initially designed their studies round the issue of
whether bilingual children started off with one system or two, by which they
meant, what type of awareness did the child have that the input they heard
around them was derived from two separate sources, and to what extent could
they reproduce this distinction? As soon as one glosses the question thus, it
becomes clear that the question is an impossible one. As De Houwer (1995)
asks, why should a bilingual child expect everything they hear to come from a
single system? Still more to the point, what would lead us to think that this is the
type of question children ask themselves?
Bilingual child language acquisition is an “automatic” process requiring no
specific or conscious effort by the child. This differentiates it from adult L2
acquisition and makes it more similar to monolingual acquisition, where it is
widely believed that children are born with a set of innate principles, such as
Universal Grammar (UG) (Chomsky, 1981; Pinker, 1994). Clearly, if this is
correct, then the capacity to learn several languages must be equally “hardwired”. One aspect of this innate capacity is the ability to identify
co-occurrences at many different levels. Thus children acquiring a pidgin as a
first language – and, according to the classic definition, thereby creating a
creole – are faced with elements which, from their parents’, or from other
adults’ points of view, emanate from two different “languages”, but which for
them are part of a single one. Adone (1994) has shown how children learning
Mauritian Creole “recreolize” their language, i.e. follow several of the predictions in Bickerton’s “Bioprogram” (1981). They thereby deviate to some extent
from the input they receive – this input being already more variable than in a
focused language situation.
Acquiring code-switching: code-switching in children (and L2 learners) 163
Children exposed to standard, focused languages, on the other hand, are
either presented with two sets of input which are systematically kept apart in
their environment (as in OPOL), or, more commonly, which are kept apart at
certain times or by certain people. They are sensitive to how those groupings
operate, be they lexical, grammatical, phonological or simply related to dialect
or register, and most acquire the capacity to produce the varieties separately as
well as to code-switch. No-one has yet systematically investigated to what
extent children brought up only with diffuse or code-switched varieties are
nonetheless able to speak the contributing varieties in a more focused or
“monolingual” fashion if required.
Much more fundamentally even than being language learners, children are
code-breakers, i.e. detectors of patterns. The fact that adults – especially highly
educated ones – call some of those patterns by a language name, be it English or
Navajo, is less of a concern to them than figuring out in what configurations it is
(a) possible and (b) common to use them. They are sensitive from an early age to
contextual factors, which suggests that, despite criticism of the concept (see
section 6.6.4), Grosjean’s modes could be useful for research on bilingual
This chapter ends with a reference to some of the points made by Tracy
(2000) in her paper “Language mixing as a challenge for Linguistics”, as these
go to the heart of the issues which have preoccupied – or should preoccupy –
researchers in this field. The first is an injunction, in analysing mixed productions, to avoid what she calls “the daisy principle” – i.e. the “picking off” of each
word or morpheme in the child’s output as belonging to one language or the
other. Even more so than with adults, many cases are ambiguous or simply not
assignable to a particular language – in any case, as we saw, the distinction is
meaningless until we are sure that translation equivalents have been acquired.
Second, Tracy points out that the fact that closed-class items are, at least in some
data-sets, the most frequently switched items (e.g. Poulisse and Bongaerts,
1994) represents a challenge to widely accepted grammatical predictions –
and common expectations – about CS.5 Third, she communicates the idea that
in “learning to mean”, children learn, among other things, that meaning may
arise from contrasts and juxtapositions, of which code-switches are one example: “It is the conflict between different representations which points to the
actual message.” Last, while formal equivalence may be maintained in language
mixing, this could be a simple outcome of the fact that “coactivated parallel
formats strengthen each other more than divergent patterns”. Appropriately
enough, she compares the code-switching child, such as the child with the
This point is not negated by the more detailed predictions of Myers-Scotton and Jake (2000)
about the different types of closed-class item and their role in production (see section
The issue here is more general, to do with the oddity – and idiosyncrasy – of such a pattern.
doll in example 1, with a linguist, testing the limits of language (Tracy,
The argument for taking full account of individual and contextual factors to
understand CS, is as strong in the case of children as in that of adults. To achieve
this, we should strive to integrate the ethnographic and sociolinguistic richness
(as provided, for example, by Zentella or McCormick) with the detailed observation of interaction which characterizes some of the classic case-studies of
bilingual child language.
What do we know about code-switching?
A fundamental argument which underlies this book is that we need to distinguish
clearly between idiolectal competence – the real locus of language – and abstract
or general conceptions of what a language is. Such general conceptions – e.g. the
notion of “correct” English, or Japanese – affect the way people speak (to a
greater or lesser extent at different times); but idiolectal competence contains
many other elements besides. When bilingual speakers code-switch, because they
are in informal mode, they are probably less affected by abstract notions of what a
language is or should be than in monolingual production. We cannot assume that
CS is always to be viewed as a combination of two entities, which, though they
can be mixed in various ways, always retain their essential character. Instead, we
should adopt a “common sense” approach, which involves looking at the product
in its own right and taking account of its internal variability in constructing
theories and definitions.
Studies described in this book highlight various characteristics of CS.
(1) CS is highly variable between communities and between individuals. This
variability gives the linguist an insight into how bilinguals’ linguistic competence is actually organized, as opposed to how the “languages” which they
officially “speak” might, in theory, mesh together. However, up to now, much
of the research has been based on an a priori approach. I have suggested that
despite the many advances in the study of bilingualism, this is connected with
a persistent, if perhaps subconscious, view of monolingualism as the natural
state of humans among many linguists. One example of this is the widespread
acceptance of the idea that it is meaningful, and even necessary, to talk of a
“base language” in bilingual speech (Gardner-Chloros and Edwards, 2004).
(2) CS is an abstraction derived by linguists from the behaviour of bi/plurilingual speakers. It takes its place within a range or continuum of manifestations of bilinguality, and should be investigated and analysed in relation to
other such manifestations, such as borrowing, convergence and creolization. Over twenty years ago, Gumperz had already moved on his definition
from the co-occurrence of two languages or varieties and described it as
“the juxtaposition within the same speech exchange of passages of speech
belonging to two grammatical systems or subsystems” (1982:56). As evidence continues to accumulate concerning intermediate and compromise
phenomena in CS, there is a need to relativize even this definition. There
is also a strong argument for the use of a more apposite term, such as
“language interaction”, to include all manifestations of language contact
without prejudging whether the varieties involved are held to belong to
discrete systems (LIPPS Group, 2000).
(3) CS shows considerable internal variation. It is particularly apt to show the
influence of sociolinguistic factors (broadly defined as all the characteristics which situate speakers within their social groupings) on speakers’
production. Second, there are significant effects related to the characteristics of the contributing varieties and the combination of more or less
closely related languages – different pairings provide different opportunities and difficulties at a linguistic, and in particular at a syntactic, level
(Muysken, 2000). From the comparisons available so far, it appears that
sociolinguistic factors are extremely powerful and can override considerations arising from apparent incompatibilities of the varieties involved.
Such a conclusion, which is likely to be further confirmed by future studies,
has far-reaching theoretical consequences.
(4) Various aspects of CS are controversial. For example, conversation analysts
argue that CS creates “local effects”, i.e. brings about meanings thanks to
its structural role in conversation, such as the use of contrasts, framing/
footing, showing cooperativeness or the lack of it (preference organization)
(Li Wei, 2002). Others, such as Myers-Scotton and Bolonyai (2001), have
argued that the varieties’ external connotations (e.g. we code/they code)
are brought to bear in code-switched conversations and add a dimension
of meaning going beyond the overt message, for which a knowledge of
factors outside the conversation itself is necessary. Although one theory is
always more attractive than two, the simultaneous operation of both types
of motivation is increasingly being recognized (Milroy and Gordon, 2003;
Sebba and Wootton, 1998). Similarly, quantitative approaches contrast with
qualitative, ethnographic descriptions, and grammatical descriptions operate largely independently of sociolinguistic ones. Such different approaches
are not necessarily incompatible, but their potential complementarity is
scarcely discussed in the literature. Linguists studying CS from different
perspectives commonly present their results as self-sufficient, and this
lack of interdisciplinarity is regrettable. This may also be true of other
fields, but that is not a reason not to address the problem in this one.
(5) CS requires an interdisciplinary approach for several important reasons. For
example, adult CS often develops out of child bilingualism, yet the development of bilingual skills in children is rarely studied in relation to the
behaviour of adults in the same community (although see the recent work
on bilingual socialization, e.g. Badequano-López and Kattan, 2007). CS is
natural, unmonitored speech par excellence, yet psycholinguistic studies
generally do not use the real CS data collected by sociolinguists and instead
investigate bilingual speech in a laboratory setting. Sociolinguists in turn
tend not to feed in to their studies the results of historical studies of language
contact, although these can throw light on the long-term effect of different
sociolinguistic factors – and so it goes on. Formal linguists, who traditionally use introspective data as evidence, are prepared to continue using
such data even in the face of the extreme variability with which CS presents
them (MacSwan, 1999; 2005). Since their enterprise is based on the reliability of native speakers’ intuitions about their mother tongue, the fact that
neither “native speaker” nor “mother tongue” can reliably be defined in
many cases of CS would seem to constitute a fairly fundamental challenge.
As Wasow (1985) somewhat roguishly put it, in basing theories on the
selection or rejection of introspectively derived sentences, they have only
their own “asterisk”.
Acknowledging fuzziness
One simple reason why CS is so often conceptualized as alternation between
discrete systems is, doubtless, that dealing with fuzziness appears less glamorous than findings based on a supposed clarity. Researchers who think “outside
the box” do not coalesce into a clear “school”, like those, for example, working
within Myers-Scotton’s Matrix Language Frame (Myers-Scotton and Jake
1995/2000). Such people, however, not only exist but have carried out highly
innovative work on CS. Many have been discussed at different points in this
book, but they are listed briefly again here, in order to illustrate the range of
arguments in favour of the “fuzzy school”, which places the complex nature of
CS in the centre stage rather than at the periphery.
Michael Clyne is one major scholar who has avoided bandwagons in favour
of a pragmatic appreciation of his data’s implications. His 1987 paper remains
one of the best-formulated challenges to the idea that CS is characterized by
universal constraints. Clyne prefers to talk in terms of some aspects of different
language combinations being “facilitators” of CS rather than in terms of constraints. His 2003 book contains a wealth of evidence on the complexity of
interlingual phenomena in relation to the languages of different immigrant
groups in Australia. He considers in detail the linguistic mechanisms of what
he terms “transversion” – basically non-insertional CS and other loans, etc.
which allow the speaker to move over from one language to another – in the
speech of these various groups, and shows how both linguistic (typological) and
sociolinguistic factors affect the linguistic outcomes. In so doing, he provides
precisely the type of comparative data which, it has been argued, is particularly
revealing of the mechanisms of CS.
Auer (1999; 2000) has stressed the difficulty, for an outsider, to determine
“the language of interaction” in a bilingual conversation – something which
the participants themselves are constantly negotiating. He describes the fluid
transition from an interactionally founded phenomenon to a structural one in
which individual switches may have no interactional significance. As we saw,
he reserves the term “code-switching” for the pragmatically motivated pole of
the continuum, pointing out that, in time, the frequency of juxtaposition of the
two varieties weakens the contextualization value of this cue.
Alvarez-Cáccamo (1998) pointed to the need to distinguish between the
linguistic material present in utterances (linguistic varieties) and the associative
mechanisms which underlie their production. He described CS as an “alloy”; in
his own data involving Galizan–Portuguese–Spanish, for example, the prosodic
patterns of one language sometimes overlap with a stretch of discourse from
another; he describes lexical, syntactic and prosodic materials from each as
being “fused into an amalgam” (p. 37).
Gafaranga and Torras (2001; 2002) and Gafaranga (2005) challenged the
idea that bilingual speakers can always be said to be speaking a particular
“language”, and pointed out that it makes more sense to think of them using a
“medium” which may be composed of several internally mixed varieties.
Switches which carry meaning for speakers and their interlocutors might not
be easily identifiable by outsiders who do not share the various hybrid “media”/
codes which form the repertoire of the particular community.
Meeuwis and Blommaert (1998) provided a “monolectal” view of CS,
emphasizing that, to the speakers, differences between “languages” may be
less salient than the differences between dialects, sociolects or speech styles.
For example, in the Zairian community in Belgium, all the national languages of
Zaire are learned as CS varieties peppered with French, which is both the
official language in the homeland and, of course, one of the official languages
in Belgium. The apparently chaotic set of combinations available can in practice
be reduced to a set of linguistic practices allowing communication through
different layers of society, but these cannot be distinguished along the lines of
the different standard languages which they include.
Rampton (1995) pointed to the need to distinguish notions of code and
speech variety. The use of particular markers, such as [d] for [δ] could represent
entire social styles. Similarly, Sebba drew attention to the minimal linguistic
changes needed to represent a switch from London English to London Jamaican
Li Wei’s work shows how individuals create personal communities which
provide them with meaningful frameworks for solving problems in their daily
lives. One of the ways in which they do this is by building social networks,
within which characteristic uses of CS develop. The structural contrast between
English and Chinese does not provide an adequate explanation of the switching
patterns, since these vary according to the group/network that people belong
to (1998a; 2000). Li Wei also points to the need for a greater appreciation of
variation between individuals in a group, confirming the idiolectal aspect of CS
which has so far been neglected (2002).
Woolard (1999) described what she called “bivalent” elements, i.e. elements
which could belong to either or both varieties. She claimed these should not be
viewed as a category to be identified, put in a box and subsequently ignored.
They are likely to be among the most revealing elements in CS speech since
they embody the underlying techniques used by bilinguals to combine their
languages. Tracy (2000) considered, from a psycholinguist’s point of view, that
the coactivation of dual systems represented by CS was nothing unusual in
itself, and that an aspect of bilingual ability was being able to suspend the
requirements of monolingual grammars. Both Romaine (1995) and Winford
(2003), who remark that neither insertional nor alternational models of CS can
cope with some of the characteristic phenomena which arise in CS (such as
operator–verb constructions and bare forms), lend support to the idea that bilinguals have meta-skills of this nature.
Boeschoten (1998) shows how the use of CS encourages a preference
for structures which show a high degree of congruence between the varieties
and emphasizes that, as a result of CS, new norms can develop at all linguistic
levels even within small (e.g. immigrant) communities. Sebba (1998) identifies
various processes through which bilingual speakers “construct equivalence”
between structures in the two varieties so as to permit CS – such equivalence is
not assumed to be a “given” within the languages themselves.
Important insights along these lines also arise in monographs describing
particular plurilingual settings. To give just two examples, Agnihotri (1987)
and McCormick (2002) both describe linguistic practices lying in a “grey area”
not identifiable with any given language.
The above are just a sample from the research which challenges the notion
of CS as a series of black and white changes from one variety to another.
By contrast, other researchers aim instead, through their descriptions of CS, to
reinforce belief in the structural integrity of linguistic systems. These include
those who promote the idea of universally valid constraints on where CS can
occur – even though, so far, no-one has succeeded in finding any; those who
assume the existence of a “Matrix Language”; and psycholinguists who
claim to be describing “code-switching” although their evidence all relates to
subjects placed in situations of categorical alternation between two standard
A third category of researchers could be said to hold an intermediate position.
They are more concerned than the first group not to topple the linguistic edifice –
the idea that linguistic production can be explained on the basis of a reasonably
consistent set of grammatical principles (leaving aside the question of how
these translate into surface rules). Yet they acknowledge a degree of variation
between speakers and within practices which defies both constraint (alternational) and matrix-based (insertional) explanations, and requires that sociolinguistic factors be taken seriously. Among these one can count Muysken (2000),
Treffers-Daller (1998), Nortier (1990) and Backus (2003). Halmari (1997)
provides an example of such a compromise. When she found violations of the
principles of UG, as instantiated by the Government constraint, which was
compatible with most of her data, and substantial interspeaker variation, she
suggested that such violations could be explained on pragmatic/sociolinguistic
grounds. Alternatively, she proposed that they could be explained by a deterioration in the speakers’ Finnish-speaking skills (1997:216–217). She referred
to the accompanying evidence of repair in their speech, such as hesitation,
pausing, backtracking, etc. as “physical symptoms of something gone awry”
(p. 215).
To some extent at least, these different approaches can be correlated with
the different types of data on which these conclusions are based. Muysken
(2000) showed that grammatical models of CS could be related to characteristics of the corpora which had inspired them (see section But the
authors mentioned above are all aware of variations in CS patterns going well
beyond their own personal data-sets. So their convictions must also be
influenced by their linguistic background and training. In the same way, the
ideas in this book show indebtedness to the model provided by Le Page and
Tabouret-Keller (both their 1985 book and more generally to its authors).
There may also be a question of temperament at play – psychologists have
pointed out the differences between those who like to work towards a single
“right answer” and those who prefer problems where there are many possible
“right answers” (Hudson, 1966).
The story so far
Regardless of our sympathies with different linguistic traditions, we should be
able to agree on what we now know, with reasonable certainty, about CS, and
what remains to be investigated. As a contribution to future research, a short
summary of the previous chapters, followed by an attempt to integrate the
findings detailed there, may be useful.
(1) It is by no means easy to define CS clearly at a linguistic level – one
person’s “code-switching” is another person’s borrowing, or community
bilingual choices (Chapter 2). Methodological matters to do with selection,
transcription and analysis of the linguistic material all play a significant
role. For example, Moyer (1998) pointed out that our view of which
language is dominant will vary depending on whether we are considering
a sentence, a conversation or a whole data-set. CS can arise as a result of
a variety of bilingual configurations in the community, including both
situations of rapid shift and of relative bilingual stability. Although it can
be symptomatic of declining competence in a variety down through the
generations, its creative use (for example in rap music) can also reinforce
the vitality, and therefore extend the life, of varieties which would otherwise
be threatened.
At a linguistic level, CS can include a variety of phenomena. Confusingly,
whereas in some cases it appears to be closely tied up with linguistic
convergence between varieties (e.g. borrowing at all linguistic levels),
in other cases it can be a way to preserve linguistic distinctiveness, while
simultaneously allowing the resources of several varieties to be used. There
are clear differences between, say, pidginization or convergence and CS, but
there are also many grey areas where similar phenomena are found, and CS
is often an aspect of these other processes. CS helps us to articulate linguistic
aspects of contact situations with the social background, which Thomason
and Kaufman (1988) showed to play such an important role in language
(2) CS is found in a wide variety of situations (Chapters 3 and 4). It can develop
into different forms within a community. Some of it may reflect an imbalance in the competence of different generations, but other instances embody
different patterns of identification with the home country, the host country
and the community itself (Chapter 3). CS can be studied – and can
be relevant – both at a macrolinguistic and a microlinguistic level. At the
macrolinguistic level, it is closely related to questions of diglossia and
interwoven with the differential values of the different varieties in the
linguistic market-place (Heller, 1998a; 1992; Bourdieu, 1997). Both
Markedness Theory, the idea that certain varieties carry a particular significance when used in certain contexts, and Network Theory, which seeks to
explain behavioural choices with reference to the relationships and bonds
which people enter into, are helpful in explaining certain instances of CS.
Gumperz’s (1982) distinction between situational CS, which arises when
the setting, topic or the participants change, and conversational CS, where
all those factors are unchanged, is still useful, though the two often coexist within the same groups and even within the same conversations.
Comparisons between communities provide the best opportunity for
assessing the relative significance of the different contributing factors.
In Chapter 4, the role of CS in structuring conversation was discussed.
This goes beyond the use of connotations and the association of different
varieties with different spheres of life, as for example when one variety is
associated with the in-group and the home, and the other with the world of
officialdom outside. Even where such connotations are relevant, speakers
may use the contrast between the varieties in their repertoire as a discoursestructuring device in itself. CS marks various conversational moves, such as
asides and quotations – the fact that the latter are not always in the language
which the original speaker used is particularly instructive. Using a different
language to that of one’s interlocutor may be a sign of rebellion or at least
disagreement. Such “verbal action” may be doubly marked, for example by
pausing as well as switching languages, or by other linguistic or paralinguistic means also available to monolinguals. CS also has functions within
Speech and Language Accommodation. It allows various “cocktails” to be
produced which may constitute a compromise between the interlocutors’
differing linguistic competences, preferences and desires to converge to, or
diverge from, one another.
Despite the many uses of CS for the bilingual speaker, studies of attitudes
towards it generally show it to be a stigmatized form of speech. This is,
however, less true among the younger generations, and a positive fashion
for CS has even be identified in some groups (Rampton, 1995; GardnerChloros, McEntee-Atalianis and Finnis, 2005). Finally, although CS is a
non-standard form of speech, it fails to show the traditional gender pattern
with respect to non-standardness, i.e. it is not systematically more frequent
in men’s speech than in women’s. Gender differences in the use of CS
depend on the functions of CS in the particular community, as well as on the
many factors which affect the use of the contributing varieties. Although
we can identify “communities of practice” and sub-groups within communities with respect to the use of CS, these are rarely simply definable along
gender lines.
(3) At a grammatical level, there is evidence of an identifiable relationship
between the grammatical patterns found within CS discourse and the
characteristics of the two contributing languages. It is also clear that CS
patterns vary according to a number of factors in the non-linguistic environment. There is no evidence that CS between any particular pair of languages
is impossible (contrary to the proposals in Sankoff and Mainville, 1986), or
that either linguistic or sociolinguistic parameters constitute any sort of
absolute constraint on what bilinguals can do. Within given language
combinations and given sets of sociolinguistic circumstances, a variety of
patterns can be found, sometimes right down to the level of idiolectal
Both insertional and alternational models of the grammar of CS seek to
describe data in terms of the interaction of two discrete systems of
linguistic rules. The notion of constraints and the notion of base or
Matrix Language imply that these systems are equated with an “external”
notion of what a language is, e.g. a variety which, if it does not have a
name such as Japanese, Russian or Swahili, is at least identifiable and
reasonably consistent. It implies that a whole bundle of features, grammatical, lexical, phonological, etc. are stored and employed as an integrated unit.
The alternative view, which is espoused here, is that individuals construct
their own systems from the input and the models to which they are exposed.
These systems overlap, but do not necessarily coincide, with officially
designated “languages” or even with varieties used by others in their
community. The notion of idiolect has so far rarely been applied specifically
to bilinguals, yet it is just as relevant as it is to monolinguals. The number
of combinations across different linguistic levels which poly-idiolectal
competence permits is potentially huge, though it is held in check by the
same urges to identify with particular groups which encourage focusing in
monolingual speakers, and perhaps by some universal processing constraints (both productive and receptive) relating to the rate of switching
between relatively stable sub-systems. To describe CS as if it were made
up of the combination of two systems outside the individual is to fail to
take account of its highly variable and personal nature. In particular the
notion of base or Matrix Language – even a composite one – has been put
in question. Although in some cases individuals may be simply grafting
elements from one language onto the underlying structures of another,
the combinations which they create are their own and cannot necessarily
be predicted by studying the potential interaction of two externally
defined systems.
(4) Psycholinguistic research on bilinguals often relies on a restricted definition
of bilinguality (Chapter 6). The subjects of experiments are often “elite
bilinguals”, i.e. students whose mother tongue is a standard language and
who have learned another standard language in an academic setting.
Moreover, the experiments used often encourage an awareness of the
need to separate languages. Psycholinguistic models of language production were originally based on monolinguals, and those which seek to
account for bilingual production often represent adaptations of the original
ones. The question of CS itself is quite rarely addressed directly in this
literature, although several topics studied by psycholinguists interested in
bilingualism are relevant. These include bilingual aphasia, where the linguistic recovery patterns of patients having suffered a tumour or lesion help
to reveal how their different languages are stored and separated, and the
localization of different languages in the brain, including the issue of whether
both languages are lateralized in the same way or whether the right hemisphere comes more into play in bilinguals. Explanations of language separation based on the localization of languages in the brain have thus far proved
unsatisfactory; and no specific switching mechanism has been identified,
though there may be a locus where switching behaviour is controlled in
the brain (Meuter, 2005). The models available describe the mechanisms of
production at a level where the complex and changing nature of CS cannot
easily be accommodated.
In the absence of clear anatomical explanations for bilingual ability,
psycholinguists such as Green (1986/2000) turned to more abstract
“models” of bilingual functioning. Relying on the notions of activation
of a given language and its opposite, inhibition, Green convincingly
suggests that CS arises when neither language is inhibited or suppressed
and so words can “reach threshold” on the basis of their availability to the
speaker, their appropriateness, etc., regardless of which language they are
in. He considers that grammatical constraints play a role in production, but
the effect of sociolinguistic factors is outside his remit.
De Bot (1992/2000) adapted Levelt’s (1989) model of speech
production to cover bilingual production and, in particular, codeswitching. He argued that macro-planning, which takes place in the
“conceptualizer”, is not language-specific, but that as the “preverbal
message” advances through the various production stages, the relevant
language-specific “formulator” is activated for different parts of the utterance. The relatively abstract nature of such a model is implicitly recognised
by De Bot, who says that there is no obvious way to put the proposals
through an empirical test.
Grosjean (2000) suggested that bilinguals function within a more or less
monolingual or bilingual “mode”, depending on the context, interlocutor,
etc. By artificially manipulating the context, he was able to show that the
amount of CS produced by subjects did indeed vary accordingly.
However, this does not tell us how these subjects would behave in real
situations where the relevant factors were both more varied and beyond
anyone’s control. Grosjean also attempted to measure the “base-language
effect”, or carry-over from one language – supposedly dominant – to the
other. Placing bilinguals in mixed-language conditions could not be said to
have a clear effect overall – in some cases it seemed to increase the
difficulty of their task, in others to make it easier. The lack of clear results
indicates that the base language concept needs further refinement, especially since it fails to account for some of the experimental evidence
(Dijkstra and van Hell, 2003) – though it could potentially be combined
with the sociolinguistic notions of accommodation and audience design.
All in all, the psycholinguistic evidence (including the evidence from the
newer brain scanning techniques) casts only limited light on natural CS. It
is suggested that in future, psycholinguists might think of ways of making
use of the abundant naturally collected data on CS rather than relying
purely on experimentation.
(5i) Research on children’s CS and that of adult language learners has
been, on the whole, contingent on a range of other relevant, but slightly
tangential concerns. As regards bilingual children, there has been
ongoing controversy as to whether they have a single linguistic system
at the outset which gradually differentiates and becomes dual, or whether
they can keep the environmental varieties distinct from the outset, at the
receptive and the productive level. Many experiments have shown that,
from the earliest age, children are able to distinguish, at the receptive
level, between linguistic features associated with different languages.
The productive level is harder to study. While it is unsurprising to find
that young children’s CS does not serve the full range of conversational
and other functions which are found in sophisticated adult CS, there is
evidence that even very young children can address speakers of different
languages in their immediate environment in the appropriate “language”.
But the “mixtures” which they produce require careful interpretation.
For example, Deuchar and Quay (2000) point out that it is pointless
to describe the child as mixing elements from two systems before they
have acquired the full set of translation equivalents – in fact the whole
notion of separate language systems may be quite inappropriate (De
Houwer 2009). CS as such in children develops in an individual way,
depending partly on parental and other input and the tolerance for it in the
(5ii) Relatively little work has been done so far on CS by second language
learners. It has been shown, however, that learners use words from their
L1 to fill lexical gaps in their target language when this does not render
them incomprehensible to their interlocutors. In practice it is not always
easy to draw a line between such CS born of necessity and more discourseoriented CS, which develops as soon as a greater level of fluency is
We should retain from all these findings regarding CS that the variety and
multifariousness which has been described is its outstanding feature, not an
incidental aspect. It is precisely because it can be used to fulfil so many different
conversational needs that it is so widely exploited. Therefore, in trying to
analyse the ways in which people alternate between varieties, we are gaining
some understanding of a whole range of characteristics of speaking itself. From
the psycholinguistic work we have looked at, it becomes clearer why studying
utterances isolated from their context is likely to lead to an incomplete view of
their characteristics and meaning. It will never be possible to place CS under a
microscope and so it will never satisfy the requirements for the object of a
scientific study in the narrow sense. Equally, it will never be possible to capture it
through a grammatical or a pragmatic approach alone. Dewaele (2005) has
argued cogently that the same is true of L2 acquisition, and has similarly called
for a “triangulated” approach combining qualitative and quantitative methods.
Psycholinguistic research could usefully be extended to take in questions where
CS is studied in its own right as opposed to being studied as a by-product
of bilinguality, e.g. through such questions as the relationship between CS and
literacy (Gardner-Chloros, 2004), or between spoken and sign language CS
(Berent, 2004).
Code-switching as an instance of a more wide-ranging
“switching” skill
As we saw in the Introduction, the term “code-switching” originated in the
world of electronics. Applying the term to human behaviour, we should bear
in mind that many of the things we do with language are instances of more
general cognitive abilities. We can make people laugh, express anger, joy and
other emotions both through linguistic and non-linguistic means. The ability to
switch between different sets of behaviour also has a more general application.
For example, Benoit, Randolph, Dualap and Johnson (2003) applied the term
“code-switching” to alternation between different sets of values. Young drug
dealers in New York are described as alternating between the culture they were
brought up in – their parents were drug users and had often been in trouble with
the law – and a code of conduct which involved strictly rejecting the use of
all drugs for themselves, except marijuana, in the interests of educational and
professional success. Von Raffler Engel (1976) looked at CS in relation to
switching between gesture-systems, but found no one-to-one correspondence
between language used and gestures.
Plurilingual CS is a particular instance of a more general ability; in particular,
what people do through CS is also done by a variety of means in monolingual
discourse. Stroud’s term “double-voicedness” to describe CS was discussed
in section 4.2.4. Some authors have gone as far as to argue that all speech is
made up, in a broad sense, of quotation. Cheshire (2004) showed how much
of spoken language is made up of unanalysed “chunks”, as did Backus in a
bilingual context (2003). Bakhtin (1986:87) writes that it is the concurrence of
word and context which creates “the spark of expression”. “These words of
others carry with them their own expression, their own evaluative tone, which
we assimilate, re-work and re-accentuate” (p. 89). Tannen shows how much
of what we say is in fact “constructed dialogue”. She claims that the term
“reported speech” is misleading as (1) most of what is referred to thus has never
in fact been spoken, and (2) even if someone’s else’s discourse is reported word
for word, those words exist “primarily, if not only, as an element of the
reporting context” (1989:101). Mayes points out that “Many direct quotes
are not based on previous utterances at all” (1990:331), and Coulmas (1986),
also writing on reported speech, concludes that there is no clear-cut distinction between direct and indirect reported speech. He identifies a hybrid form
which he calls “quasi-direct speech”. In CS, even the use of a single word in
another variety can introduce the sense of quotation, of momentarily invoking another persona or identity. We saw this clearly in the case of the young
Greek Cypriot women whose use of Greek “legitimized” humorous or direct
interventions, which were not part of their normal English-speaking selfpresentation.
Along with the use of language, it is human beings’ flexibility/adaptability
which distinguishes them from other members of the animal world. They
can live in very hot climates and very cold ones. They can be meat-eaters or
vegetarians. The ability of many of them to alternate between different
linguistic varieties at the drop of a hat, and to make use of several simultaneously when it suits them to do so, is part of the adaptability which has
ensured their survival.
Some directions for future research
Several priorities for future research have emerged during the course of this
From the point of view of grammar and sociolinguistics, the priority is to
set up comparative studies which will allow us to gauge the effect of the
different types of variable which have been found to affect CS, both
linguistic and sociolinguistic. As Deuchar wrote with respect to child
language, rather than asking whether the child has one system or two, one
should look at the processes of differentiation – how and when they occur.
A methodological framework for analysing data comparatively has been
proposed in the LIPPS coding manual (2000), which is outlined in the
At a psycholinguistic level, there is a need to study CS as it occurs in more
realistic contexts than the rarified atmosphere of the laboratory. Sociolinguists
are aware of the Observer’s Paradox, but psycholinguists seem less aware of
the paradox involved in studying unregulated, spontaneous and creative social
behaviour in a controlled setting with limited parameters. In so doing one may
be studying something relevant to CS, but not CS itself. One suggestion as
to how this might be done is to carry out analyses of authentic recordings of
spontaneous CS, for example by measuring pausing, VOT and other aspects of
speech production in CS and monolingual conditions. Any loss in scientific
precision would be more than made up for by increased relevance.
Phonological studies of CS are under-represented in the CS literature
(Botero, Bullock, Davis and Toribio, 2004). Yet borrowing and CS can
occur at all linguistic levels. Clyne (2003) describes phonological aspects
of “transversion” as being intimately bound up with lexical, structural, etc.
ones. The fact that “accent” in many people’s speech carries over from one
language to another is an integrating factor, lending unity or “equivalence” to
elements from what could otherwise be termed separate languages. The role
of idiolectal “accent” is rarely mentioned as a factor when discussing whether
or not items are phonologically integrated, yet this in turn has been used as a
criterion to distinguish borrowing and CS – another example of the individual being left out of the account. Woolard’s (1999) injunction to pay more
attention to “bridge” or “bivalent” elements, rather than simply cataloguing
them and then leaving them outside the main analytical framework, is an
important one.
The past and future of code-switching
In a chapter of his book Imagined Communities, about the origins of national
consciousness and nationalism, Anderson (1991) remarks that the development
of printing gradually led to a marginalization of those dialects which did not
correspond with the printed varieties. In the nineteenth century, this led to a
complex relationship between national “print-languages”, national consciousness and nation-states (1991:48–49). The idea of “One Language One Nation”
arose at that time, along with notions of linguistic purity and exclusiveness. This
ideal contrasts with the laissez faire in earlier historical periods, when several
varieties could be used in the same stretch of discourse – even in written records
and literature destined for serious purposes. CS was common in writing as well
as speech in various historical periods; it is found in the Acts of the Apostles,
in medieval literature and political texts, in the writings of Luther and the songs
of the troubadours.
At the beginning of the twenty-first century, we are still deeply affected, at
institutional and other levels, and particularly in the educational sphere, by ideas
originating in the nineteenth, and in particular by the notion of monolingualism.
Furthermore, economic transformations in the twentieth century, notably the
rise to dominance on the world scene of a basically monolingual (or assimilative) country, the United States, combined to encourage a negative mindset
towards hybridity, “half-castes” and cultural fusion. Fishman’s work (1968) on
the correlation between multilingualism and poor economic status represents a
confirmation, for linguists, of this unpalatable truth.
However there are signs that the tide is changing. There has been an
increasing empowerment of certain multicultural and multilingual groups,
e.g. Hispanics in the USA. There are fashions for what is known as “fusion”
in many areas of life. Westerners have adopted elements of traditional dress
and decoration from far-flung countries instead of dismissing them, as they
would once, as folklore; Italians make their own homegrown version of sushi
and rappers rap in multiple codes. “Crossing”, as described by Rampton
(1995), is the linguistic embodiment of this trend, which looks set to continue.
We should not, however, trivialize CS by considering it merely a matter of
fashion. Perhaps particularly in young people, it embodies a range of conflicts,
ideologies, oppositions to authority and positionings with regard to gender and
ethnicity which should give us more wide-ranging food for thought (Pujolar,
Even the educational establishment has not remained unmoved. In Chapter 3
we saw that schools in many countries of the world tolerate, and sometimes even
encourage CS. Amherst College, in Massachusetts, has a professor of Espanglish,
the Spanish–English mixed code of Hispanic immigrants, who has written a
dictionary of Espanglish of six thousand words. Students are set translations.
The first two sentences of a translation of Don Quixote into Espanglish run as
Example 1
In a placete de la Mancha of which nombre non quiero remembrearme, vivía, not so
long ago, uno de eses gentlemen who always tienen una lanza in the rack, í una buckler
antigua, a skinny caballo y un greyhound para el chase. A cazuela with más beef than
mutón, carne choppeada para la dinner, un omlet pa los sábados, lentil pa los viernes,
y algún pigeon como delicacy especial pa los domingos, consumían très cuarers de su
‘In a place in La Mancha the name of which I do not wish to remember, there lived, not
so long ago, one of those gentlemen who always have a spear in the rack, and an antique
buckler, a skinny horse and a greyhound for the chase. A casserole with more beef than
mutton, meat chopped up for dinner, an omelette for Saturdays, lentils on Fridays and
some pigeon as a special delicacy for Sundays, consumed three quarters of his income’
(my translation). The words in bold are English words adapted to Spanish morphology.
(Mangruel, 2004:178)
As defined here, Espanglish is a code-switched variety incorporating English
loans adapted to Spanish morphology.
Naturally, the growing strength of Hispanics in the USA is unusual, and
only a small proportion of linguistic minorities is gaining ground to that extent.
But it does seem likely that CS will become increasingly acceptable in many
contexts. Many would argue that developments like that exemplified above,
the acceptance and institutionalization of CS, are retrograde and even harmful.
We should be clear that such judgements are political, aesthetic and cultural –
not linguistic. The weight of evidence is that bilingualism is beneficial rather
than harmful at the individual level, and that learning to code-switch is no
different from learning any of the other “rules of use” of any other variety.
Nor can any easy conclusions be drawn as to whether such developments will
precipitate the impoverishment of particular languages. They are more likely
to be a symptom than a cause, and in many cases the alternative may be total
death. CS may even keep some languages alive, perhaps for long enough so
that the next generation is intrigued enough to revive and teach them.
Zentella writes of her Spanish–English code-switching subjects on el bloque
in New York: “They were competent in Spanish, but insisting that English
has not influenced the community’s Spanish is akin to maintaining that the
experience of being born and/or raised in the US has had no impact on
second generation Puerto Ricans, as if the young passively inherited instead
of actively created their culture” [my italics] (1997:271).
Code-switching can be likened to a linguistic barometer, and we should
continue to investigate the pressures and fluctuations which it records and
embodies, which arise in different contexts all over the world. It is a goldmine
for linguists, because it highlights so many important questions about the
relationship of languages to language. A full understanding of CS requires us
to take account of all aspects of linguistic enquiry, including grammar, sociolinguistics, psycholinguistics and more. Its study should contribute to putting
interdisciplinarity on the linguists’ map at last, and in the process, let us hope, to
a small and much-needed methodological revolution.
Coding and analysing multilingual data:
the LIDES Project1
Over the last few decades, data on bilingual and multilingual talk has been
collected through projects, large and small, in many countries and involving
many different languages and dialects. It has become a source of frustration for
many who work in this area that there is no basic standard for transcribing data
of this kind, nor any central resource to enable researchers to share their data
with each other. Meanwhile, researchers in some fields have begun to take
advantage of new technological developments to enable data to be shared.
Researchers in language acquisition, for example, have both standard ways of
transcribing and coding data, and international databases to which they can
contribute and on which they can draw for comparative data.
The purpose of the LIPPS initiative is to provide support and propose a system
for transcribing and coding bilingual data that takes into account research questions specifically related to multilingual communities and individuals.
Researchers in the field may benefit from the LIPPS initiative in two ways:
(1) By following the step-by-step guidelines to carry out the transcription and
coding which make it possible to use already existing computer-based
analytical tools.
(2) The existence of a set of basic standards for transcribing and coding
bilingual data should encourage research on language interaction from an
LIDES = Language Interaction Data Exchange System. LIPPS = Language Interaction in
Plurilingual and Plurilectal Speakers. This Appendix is an abridged version of the chapter
“Coding and Analysing Multilingual data: the LIDES Project” by Penelope Gardner-Chloros,
Melissa Moyer and Mark Sebba (2007), which was itself derived from a longer description of the
system in a special issue of the International Journal of Bilingualism, 4, 2, June 2000. However,
this did not include various developments to the system which have been made since then. The
authors of that volume were the LIPPS Group and included (in alphabetical order) Ruthanna
Barnett, Eva Codó, Eva Eppler, Montse Forcadell, Penelope Gardner-Chloros, Roeland van Hout,
Melissa Moyer, Maria Carme Torras, Maria Teresa Turell, Mark Sebba, Marianne Starren and
Sietse Wensing. Permission from Jill Lake of Palgrave publishers to abridge the chapter here is
gratefully acknowledged, as is permission from the editor of IJB, Li Wei, to reprint the glossary
from the special issue of IJB.
interdisciplinary perspective. Thus, a special effort is made to cater for the
needs of researchers working in very different disciplines within the field.
The proposals made here are equally intended to help those who are
interested in quantitative and qualitative research. Useful guidelines are
provided so that solutions may be found for problems that may arise in the
processes of transcription and coding. In addition, some user-friendly
computational tools are discussed which provide support in exploring and
analysing language interaction data.
We have adopted the term language interaction rather than the more commonly used terms “code-switching” and “language contact” in order to include
all manifestations of language contact, whether or not the varieties under study
are held to belong to two discrete systems.
The goals of LIDES
In spite of the high level of interest in language contact, no coordinated system
has yet been developed for researchers to make use of one another’s data. The
data is only available, if at all, in widely disparate forms, and coding and
transcription practices vary significantly. It is even rarer that anyone except
the original researcher gains access to the original data (usually audio
The LIDES project is modelled on the CHILDES enterprise established by
Brian MacWhinney. It consists of a set of standards and computational tools for
analysing language acquisition data, and a collection of data-sets contributed by
diverse researchers (MacWhinney, 1995; MacWhinney & Snow, 1990), and
shows the advantages of being able to access a database in research fields where
data on spoken, spontaneous language is essential. Each researcher works
independently on his/her own data-set, but a common set of overall goals is
kept in mind.
The reasons why it is desirable to achieve a coordinated approach go beyond
the advantages of simple data sharing. What researchers typically want to know
about patterns of language interaction is to what extent these patterns are
dictated by the particular language combination and/or the context and circumstances which are relevant in their study, and to what extent they are universal
or at least common to similar language sets or similar combinations of sociolinguistic circumstances. For example, one major strand of research on codeswitching focuses on grammatical constraints on where a switch can occur
within the sentence. Time after time, constraints proposed on the basis of
one data-set (and often put forward as potentially universal) have been
disproved when new data-sets have emerged (Muysken, 2000). Furthermore
it is not possible, without making comparisons of the kind we propose, to
establish the relative role of linguistic features as such and sociolinguistic,
Coding and analysing multilingual data
psycholinguistic and/or contextual factors in the language interaction patterns
that are observed. Both of these are fundamental problems with approaches
based on a single data-set.
In this section, we discuss the rationale for basing LIDES on the CHILDES
system and the CHAT coding protocol. Section 3 is devoted to the basic steps for
preparing language interaction data for analysis. Special attention is dedicated to
the steps involved in the preparation and organization of the data and to the
minimal requirements for transcription and tagging. Section 4 runs through
the essential elements of a CHAT data file. Section 5 takes the reader through
the transcription process step by step. In Section 6 we mention some ways
of coding data for specific research interests. Section 7 describes some of the
programs that automate data analysis. Finally, Section 8 contains practical information about contact addresses and obtaining the LIDES programs and datasets.
Why use CHILDES?
CHILDES has been successfully used for fifteen years and is equipped with an
institutional support base,2 specific detailed guidelines for transcribing and
coding data (the CHILDES coding manual) within an existing format (CHAT)
and a set of software, the CLAN programs, which researchers can use to carry
out a large range of automated analyses of the data in the database. The
programs for analysis of bilingual data can be obtained by contacting Brian
On the other hand, CHILDES was not designed for adult, multilingual,
speech data. It was set up to provide a database of conversations involving
mainly monolingual adult–child interactions. Therefore, the CHAT format was
not initially the most appropriate one for the type of data researchers in this area
were collecting, and the CLAN programs were not designed to answer the type
of questions that researchers working in the field of multilingualism would want
to ask. However, the CHAT coding scheme and the CLAN tools in CHILDES
were adopted precisely because they are open to further elaborations and
additions. In fact, some adaptations, necessary to cope with multilingual data,
can already be found in CHILDES and others have been developed by the
LIPPS group. Even greater possibilities are offered by the development of a
framework based on XML for CHILDES and TALKBANK (MacWhinney,
2007). The CHILDES database contains data from many different languages,
and the way transcription problems have been solved for these different languages can be of help when transcribing language interaction data.
The CHILDES system is maintained by Brian MacWhinney at Carnegie Mellon University,
Pittsburgh, USA and the University of Antwerp, Belgium; email: [email protected]; internet:
More important in this respect, though, are the bilingual data already available. A separate chapter of the CHILDES manual (MacWhinney, 1995) is
devoted to data available on bilingual acquisition, which is becoming an
important field of research. In the CHILDES system different types of information are coded on a series of “dependent tiers” linked to the “main tier” which
carries the transcription. A separate dependent tier is proposed to code information on the language of the utterance on the main tier (see below), the language
of the preceding speaker and the dominant language of the speaker (see
MacWhinney, 1995: 63; De Houwer, 1990, for a good example of this type of
system). There is also a separate subsection on code-switching in the CHILDES
manual (MacWhinney, 1995: subsection 9.4) in which some useful coding
options are proposed.
Another point in favour of the CHILDES system is the formal way in
which the system is set up. One development is an interface between CHAT
and XML formats. XML is a markup metalanguage of the World Wide Web
with powerful tools for analysing data on the web. It is a standard way of
marking up texts (both spoken and written), which is independent of any
word processor or computer system. One of the present tasks of the LIPPS
group is to offer a universal set of codes for language interaction data in
XML. The new system will also offer user-friendly interfaces which will
allow researchers to encode numerous different scripts and have a choice of
how to present the transcribed and encoded data in print and on the computer
screen. A program for converting CHAT files to XML can be downloaded
from the CHILDES web page. This prospect matches the view of
MacWhinney (1995:437): “As our work in database development proceeds,
we want to think in terms of a more general database of all the varieties of
spoken human language.”
Another recent innovation to CHILDES that has also been incorporated
by LIDES is the adoption of the Unicode encoding system. CLAN programs
presently recognize this new system. Unicode is important for research on
multilingual data because it allows language interaction researchers to use
their computer keyboard to represent a character from any language including Arabic, Chinese or IPA that have different script types. A further
advantage to using Unicode is that it permits researchers working in the
field of discourse and conversation analysis to use the CA-CLAN programs
developed to analyse utterances, turns, overlaps as well as other conversation phenomena used in the transcription conventions put forth by Atkinson
and Heritage (1984).
A far-reaching expansion that has also been developed in recent years is the
linking of original digitized audio and digitized video–audio recordings to
transcribed files. Linking can also be done at the same time one carries out
the transcription. New avenues for spoken language analysis and possibilities
Coding and analysing multilingual data
for checking and revising transcriptions can easily be carried out. More detailed
information on linking is provided with the CHILDES programs. A computer
program called PRAAT, with which it is possible to analyse, synthesize and
manipulate speech, has been developed by Paul Boersma and David Weenik.
This program is especially useful for splicing short audio files into a single large
file. The CHILDES project is currently developing further CLAN support for
this program.
Why use CHAT?
The existing CHAT transcribing and coding conventions are very flexible, making it possible for the researcher to reflect many kinds of phenomena that occur in
natural speech data. We give some examples of existing CHAT transcription and
coding options below. Furthermore, you can add any type of code you want, as
long as you use it consistently and define it properly, which is what you would
have to do anyway, be it in a more traditional format or the CHAT format.
You can see from examples one and two that transcribing in CHAT is not that
different from the traditional way of transcribing, and that the basic transcribing
and coding conventions are not that difficult.
(1) A “traditional” transcription (Moyer, 1992)
The participants in this conversation are Yvonne and Natalie, both housewives.
The languages used are Spanish and English. English is given in plain typeface,
Spanish in italic typeface.
yvo: Excuse me could we have two coffees and some scones please?
nat: Yvonne para mí no vayas a pedir scones de esos
Yvonne for me not go to ask scones of these
que ahora me estoy tratando de controlar un poquito antes de Pascua
that now me are trying to control a little-bit before of Christmas
“Yvonne, don’t order these scones for me because now I am
trying not to put on weight before Christmas.”
yvo: sí Christmas ya está round the corner mujer
yes Christmas already is round the corner woman
“Mind you, Christmas is already round the corner.”
In this traditional transcription we can see:
there is a short introduction giving details of speakers and languages used;
each speaker’s turn is put in a separate paragraph following an indication of
who is speaking;
normal and italic fonts are used to indicate the language of each word/phrase;
the literal gloss for each word is placed on the line beneath; and
there is a free translation of the conversation provided.
In the following example the same data is given in CHAT format:
(2) a CHAT transcription (Moyer, 1992)
N.B. Changes are constantly being made to the system. At the time of writing, a
new presentation has just been adopted for the bilingual data in Talkbank,
whereby only one of the languages is overtly coded. The appropriacy of this
for code-switched data in which the dominant language is constantly shifting,
or indeterminate, is under discussion in the LIPPS Group.
@Participants: YVO housewife1, NAT housewife2
@Languages: English (1), Spanish (2)
*YVO: [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected]?
[email protected] [email protected] mí@2 [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
Yvonne for me not go to ask scones of these that now me are trying to
control a little-bit before of Christmas.
Yvonne, don’t order these scones for me because now I am trying not to
put on weight before Christmas.
*YVO: sí@2 [email protected] [email protected] está@2 [email protected] [email protected] [email protected]@2.
%glo: yes Christmas already is round the corner woman
%tra: mind you, Christmas is already round the corner
Transcription (2) uses the coding system recommended by LIDES, and as you can
see, the data looks somewhat different. However, the same information is present:
there is a set of “headers” giving details of speakers and languages used
(@Participants, @Languages);
each utterance is put on a separate line (the “main tier”) following an
indication of who is speaking;
language tags (@1 and @2) are used to indicate the language of each word/
there is a separate line or “dependent tier” following the main tier for the
literal gloss, the %glo tier; and
there is a dependent tier for a free translation of each utterance, the %tra tier.
From the above examples, it is clear that the CHAT transcription is actually
very similar to the traditional one. Some extra work is inevitable, for example,
tagging each word with a language code, though word processors or “editors”
are currently being developed to help do this task automatically.3 However,
once you have done this work, the benefits are substantial. To name a few, you
can add as many dependent tiers as you need. You can create additional headers
For example, you can use a macro to build up a lexicon of words to be coded by language.
Coding and analysing multilingual data
which provide information such as the socio-economic status or age of the
participants, the name of the transcriber and the date of the recording. You can
create additional dependent tiers on which you can provide a gloss (%glo), a
translation (%tra), or code grammatical, pragmatic or other information. You
can devise codes to label problematic words which do not clearly belong to one
language or another, such as in when dealing with English, Dutch and a range of
Germanic languages. Once your transcription is complete, you can print it,
selecting just those tiers you want to appear on the page, from among all those
you have included in your transcription.
An enormous benefit, obvious to every researcher who has done frequency
counts using pen and paper on a relatively large set of data, is the fact that you
can use the CLAN programs to search your data for patterns or provide certain
statistics. This is because all LIDES transcriptions can use either Unicode
characters and a common set of transcription and coding guidelines. The latter
also makes it easier to exchange and compare data with other researchers, inside
or outside the LIDES database.
Finally, although the LIDES system recommends the use of a transcription
system based on CHAT to transcribe data, existing data in different formats is
not excluded from LIDES. Such data can still be used productively by other
language interaction researchers if included in the database. One of our aims is
to develop adequate tools to convert these data sets into the transcription format
utilized by the CLAN automatic analysis programs.
The basics of transcription
Transcription is not pre-theoretical, but is a form of interpretation of the data, as
discussed by Ochs (1979). For any set of spoken data, there is no single
“correct” method of transcription, but a variety of theory-dependent options.
The problems, both theoretical and practical, which are inherent in the transcription of monolingual discourse, are multiplied when two or more language
varieties are present. Where plurilingual data is involved, most researchers want
a transcription system which provides a way of differentiating between the
languages (or language varieties) found in the data. One solution may be to use
the conventional writing system of the languages concerned, if there is one. If
the researcher chooses a “language-neutral” alphabet such as the International
Phonetic Alphabet (IPA), they will need some method of indicating which
stretches of transcription are in which language. There will still be problems,
however. For some utterances it will not be easy to assign them to either
language: for example, fillers like er and um and word fragments of various
sorts which are not necessarily language-specific.
What to do in these cases involves theoretical decisions being made by the
individual researcher. It is not simply a matter of choosing a method. Assigning
a word, by means of some transcription convention, to either L1 or L2, may
have consequences later for the outcome of the investigation. The researcher
must have a principled basis on which to make these decisions, and may have to
take into account aspects of syntax, phonology and pragmatics. Where more
than two languages or related varieties are involved, things are even more
If researchers are to be able to establish large usable machine-readable corpora
of code language interaction data, then a standard format is needed for transcribing and coding the data that fulfils at least the following requirements:
It should allow information contained in transcriptions (for example, language labels) to be stored in a format which is not word-processor dependent.
This involves using a system-independent format such as Unicode along with
a standard markup language such as XML. This enables data transcribed by
one researcher to be compared easily with data from another, as well as
promoting a more universal software to carry out rapid searches and analyses
of data.
It should encourage researchers to include in their transcriptions, for their
own benefit as well as for that of other researchers, information which is
relevant to the analysis of language interaction data. This is desirable as a
way of helping researchers new to the transcription of language interaction
data to include features which will be useful for their own analysis at the
time, and to others who may want to use their data later for another
It should be flexible enough to accommodate the present and (as far as is
predictable) future needs of researchers in terms of what needs to be represented in a transcription. Researchers should feel in control of the transcription process rather than being forced into a straitjacket. The guidelines
have no value if researchers do not find them suitable for their own needs.
CHAT allows researchers to include all the information needed for the
analysis of their data. Since language interaction data differ according to the
language pairs involved, researchers also may ask very different questions
regarding their data and thus will want to codify different types of information
in order to carry out an analysis. CHAT has the advantage of being adaptable to
individual research needs. We give some examples below of language interaction data transcribed and coded in CHAT, including some of the proposed
LIDES adaptations.
In the following example a tier (% add) has been inserted to indicate the
addressee. From this it is possible to see that the speaker changes the language
according to the person he/she speaks to:4
It is good practice to replace
This example contains a number of non-ASCII characters (n
these with sets of Unicode symbols, making clear the mapping between these and the orthographic
characters they represent.
Coding and analysing multilingual data
(3) Catalan/Castilian: Torras (1998)
@Participants: OWN owner stall_owner, CU1 customer_one customer, CU2
customer_two customer
@Languages: Catalan (1), Castilian (2), unintelligible (0)
*OWN: digui’[email protected]
can I help?
*CU2: [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected][email protected] [email protected] [email protected] [email protected]
that xxx of the shoulder over there the smallest you have.
*OWN: [email protected] [email protected] [email protected] [email protected] [email protected]
I’ll have a look for you right now.
%add: CU2
*OWN: qué@2 [email protected] [email protected] señ[email protected]?
what are you giving me madam?
%add: CU1
*OWN: [email protected] [email protected] [email protected] [email protected]
two hundred and seventy two.
%add: CU1
*OWN: [email protected] bé@1 grà[email protected] [email protected] [email protected]
very good thanks xxx love.
%add: CU1
A more complicated example is the following. In this example, codes for speech
acts are added on the speech act coding line (%spa:) to show how a positive answer
was given in the same language as the question was (B answers her mother’s
question twice, first in English, second in Cantonese, both times using the same
language as her mother used), whereas a negative answer is given in another
language than that in which the question was put (when A asks C in English
whether he wants some spring rolls, the answer is ‘I don’t want’ in Cantonese):
(4) Cantonese / English: Milroy and Li Wei (1995:149), adapted to CHAT
@Age of B:
@Age of C:
MOT mother, DAU daughter, SON son
Cantonese (1), English (2), undetermined (0)
[email protected] [email protected] [email protected]?
<[email protected] [email protected] > [>].
<[email protected]> [<].
[email protected] [email protected]?
want some?
[email protected] [email protected]
$i:yq $i:cl
[email protected] [email protected] [email protected]
I don’t want.
[email protected] [email protected] [email protected]?
Don’t want?
[email protected] [email protected]
[email protected]
shaking head
Key to symbols
scope of phenomenon
overlap follows
overlap precedes
gloss; LIDES recommendation
speech act coding tier
gestural and proxemic activity coding tier
Speech act codes (MacWhinney, 1995:101–103):
illocutionary force code follows
yes/no question
answer in the affirmative to yes/no question
answer in the negative to yes/no question
call attention to hearer by name or by substitute exclamations.
Coding and analysing multilingual data
Other interesting features in this example are the way overlaps are transcribed
(using <>, [<], [>]), and how various non-linguistic activities are rendered in
CHAT (%act:, %gpx:).
The essentials of a CHAT data file
Once the researcher has completed transcribing and encoding the data, his/her
data set will contain three or four files: the CHAT data file, the depfile, a read me
file and possibly a depadd file.
The CHAT data file
A CHAT data file typically consists of three components: file headers, main
tiers, and dependent tiers. For all of these components the following general
requirements obtain (MacWhinney, 1995:8):
(a) every character in the file must be in the basic Unicode character set;
(b) every line must end with a carriage return (= enter).
Other requirements obtain only for a specific component, and will be given
where the component in question is discussed.
A.4.1.1 File headers File headers are lines of text which provide information
about the file and they appear at the beginning of any transcript. They all begin
with the sign @ followed by the name of the header. Some headers require
nothing more; these are the “bare” headers (for example, @Begin). Other headers
must be followed by information regarding the participants, the situation and so
on. This information is called the entry. Headers that require an entry are followed
by a colon and a tab after which the entry is given. For example:
(5) @Participants: SHO shopkeeper, RES researcher, SEA seamstress
[Besides the distinction between bare headers and headers requiring an entry,
CHAT makes a distinction between obligatory and optional file headers.
Obligatory file headers (MacWhinney, 1995:13–14) are required for the
CLAN automatic analysis programs to work. These are the following:]
The first line of any transcription file. This is a bare header.
@Participants: This header states who the participants of the conversation are. It
requires an entry which may consist of two or three parts: a
speaker ID (obligatory) which is composed of one to three
characters (letters in capitals and/or numbers), for example,
MAR, or S09; the speakers’ names (optional); and the
speakers’ role (obligatory), for example, Interviewer. A role
which is not included in the depfile must be added in the
depadd file (see Glossary and Section A.4.3). The latter two
parts may consist of multiple elements, but they must be
connected with an underscore, for example, Penelope_
Gardner_Chloros or sister_in_law
The last line of any transcription file. @End is a bare header.
There are two types of optional headers: constant and changeable. Constant
headers contain useful information that does not change throughout the file.5
They must be placed at the beginning of the file before any of the actual spoken
utterances. Because of the interaction of different languages in LIDES files, the
following two file headers are recommended to be used consistently:
@Language of
This header indicates what the main language or languages
of the transcription files are.
This header can be used to indicate the primary language(s)
of a given participant, if there is any.6
Changeable headers contain information that can change within the file, for
example, showing how the conversation can be divided into several stages each
associated with a different activity. These headers appear, then, at the point
within the file where the information changes. Besides these headers, new
headers may be created to fulfil specific needs as long as they are added to
the depadd file located in the same directory as your data files.
A.4.1.2 The main tier The actual utterances are transcribed on what is known
as the main (speaker) tier. This is the line that reproduces in written form what
each one of the participants says. Requirements for the main speaker tiers are as
follows (MacWhinney, 1995:8–9):
Main lines indicate what was actually said and begin with an asterisk*. Each
main line should contain one and only one utterance7 (but can extend over
several computer lines).
After the asterisk * on the main line comes a code (the speaker ID, see the
Participant header) for the participant who was the speaker of the utterance
being coded. After the speaker ID comes a colon and then a tab; for example,
*MAR: I’m tired.
Some headers may be considered either constant or changeable, for example, date, location, situations
and so on.
There are different views as to whether it is possible, always or even sometimes, to decide which is
the primary language, and if so how this should be done. The reader is referred to discussions
surrounding the “Matrix Language” as defined by Myers-Scotton (see for example, GardnerChloros and Edwards 2004: 117–120).
The notion of utterance may vary from one researcher to another depending on specific research
interests. LIDES purposely does not provide a single definition of utterance. Some criteria for
distinguishing an utterance can be a single word, an intonation unit, a long pause and so on.
Because of the possibility of sound linking to the transcripts the issue of where to put boundaries
on an utterance may become somewhat less arbitrary, but we do not expect it will solve the
segmentation problem completely.
Coding and analysing multilingual data
Continuations of main tiers over several computer lines begin with a tab.
Utterances must end with an utterance terminator, that is full stop (.),
exclamation mark (!) or question mark (?).
A.4.1.3 Dependent tiers Dependent tiers are lines typed below the main tier
containing codes, comments and descriptions of interest to the researcher. It is
important to have all this information on separate lines because the use of
complex codes on the main tier would make it unreadable. Though all dependent tiers are optional, we recommend having at least a gloss tier (%glo) and a tier
in which a free translation of the utterance is given (%tra). Requirements for
dependent tiers are the following (MacWhinney, 1995:8–9):
Dependent tiers typically include codes and/or commentary on what is said
and begin with the symbol %.
After the % symbol comes a three-letter code in lower case letters for the type
of dependent tier, a colon, and then a tab; for example, %glo: I am tired.
Continuations of dependent tiers over several computer lines begin with a tab.
Dependent tiers do not require ending punctuation. The researcher’s interests
determine the number and nature of dependent tiers in a transcript. In the
examples above, we saw the following dependent tiers:
%add: specifies the addressee(s) of the utterance.
%glo: gives a word-by-word or even morpheme-by-morpheme gloss of the
%spa: this tier is for coding speech acts. Any sort of speech act can be used to
describe the utterance on the main tier. A proposal for codes to be
used on this line can be found in MacWhinney 1995: Chapter 13.
%tra: tier where a free translation (in English or some other widely known
language) of the utterance is given. We propose to use %tra instead
of the %eng tier in the CHILDES manual, %tra being more language
Other useful tiers include:
%gpx: gestural and proxemic activity coding line, for example:
%gpx: shaking head
%mor: coding of morphological information for the words on the main tier.
Codes used on the %mor tier can be found in MacWhinney 1995:
Chapter 14.
In order to make it possible to exchange all the data coming from different
researchers using different computer systems (Mac, PC, UNIX and so on) and
printers, we strongly recommend that data be transcribed in Unicode format,
by using a true Unicode editor. In the CHILDES manual (1995), you can find the
description of a Unicode editor specially created for the transcription (and coding)
of CHAT files: the so-called CED editor. This allows the user to link a full
digitized audio recording of the interaction directly to the script. Furthermore, the
CED editor supports the display of non-ASCII Roman-based characters such as á,
ñ or ç, as well as non-Roman characters from Cyrillic, Japanese, Chinese and
other languages. In all cases, CED displays these fonts correctly, but the underlying file is saved in Unicode characters. The CED editor is included in the CLAN
package, which can be downloaded from the internet site (see Section A.8.3).
The readme file
All data sets should be accompanied by a readme document (00readme.doc),
which is aimed at providing general information about them. Information which
is specific about a particular file in this set should not be included here but in the
header of the specific file. Below is a checklist of information that the readme
file should specify:
researcher (name, institution, history of the project, etc.);
characteristics of the community to which informants belong;
informants (age, sex, class);
sampling techniques;
number of hours recorded;
number of hours transcribed;
special transcription and coding practices;
interaction type;
working definitions used to identify given language interaction phenomena;
warnings about limitations on the use of the data (that is, what has been
transcribed and coded and what has not);
list of files;
instruments used for data collection;
changes made in the 00depaddfile.
This list is not meant to be exhaustive. Contributors may add whatever they
consider useful for future users. It may be helpful to comment on a particular
aspect of a language or variety so that readers unfamiliar with it have a better
understanding of the data. For example, a researcher dealing with Swahili data
may want to provide an outline of the null prefix system of this language because
he/she thinks this is relevant to the study of his/her corpus.
depfile and depadd files
These are files which list the codes used in the data file and are needed by the
CLAN programs for them to be able to check your transcription and carry out
analyses. Depfile.cut is a standard file that is delivered with the CLAN programs. The 00depadd file is a file you make yourself. For more information on
Coding and analysing multilingual data
the contents of the CHILDES depfile, and their meaning, see MacWhinney
For analysing language interaction data, it is possible to extend the CHAT
depfile by creating an additional file, a “depadd file” called “00depadd.cut”,
which you should place into the same directory as the files being checked. All
new codes must be included in this corpus-specific LIDES depadd file: new
header codes, new dependent tiers and the possible strings that may occur on
those tiers. The LIDES depadd file (changed only if necessary) should also
remain with the data files as a form of documentation of the particular divergence from the standard CHAT depfile and the standard LIDES depadd file. In
the readme file, you should describe any changes you made in the standard
LIDES depadd file.
A step-by-step outline of the minimal transcription process
This section gives a step-by-step outline of the basic information a data file should
contain in order to enable researchers to carry out the analysis of switches
between languages. You can find the details of various steps in the sections
referred to at these steps. There are eight steps (data from Moyer, 1992):
Step 1. Transcription
Y: Excuse me, could we have two coffees and some scones, please?
N: Yvonne, Para mí no vayas a pedir scones de esos que ahora me estoy
tratando de controlar un poquito antes de Pascua.
Y: Sí Christmas ya está round the corner, mujer.
Step 2. Obligatory File Format
@Participants: YVO housewife1, NAT housewife2
*YVO: Excuse me, could we have two coffees and some scones, please?
*NAT: Yvonne, para mí no vayas a pedir scones de esos que ahora me estoy
tratando de controlar un poquito antes de Pascua.
*YVO: Sí Christmas ya está round the corner, mujer.
*YVO: Yo ya no hago dieta hasta por lo menos enero, febrero y eso con suerte
Step 3. Run the CHECK programs
In order to check the overall structure, the CHILDES system provides special
CHECK programs which runs twice over the files. In the first pass, CHECK
searches for errors by comparing the files to the prescribed format. If errors
are found (for example, no @Begin and @End markers at the beginning and
end of files) the offending line is printed, followed by a description of the
problem. On the second pass it checks if the used symbols and codes are
declared in either “depfile.cut”, or in “00depadd.cut”.
Run the CHECK programs now to check if the basic format is correct.
Step 4. Language tagging
@Participants: YVO housewife1, NAT housewife2
@Languages: English (1), Spanish (2)
*YVO: [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] and
@1 [email protected] [email protected] [email protected]?
*NAT: [email protected] [email protected] mí@2 [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
*YVO: Sí@2 [email protected] [email protected] está@2 [email protected] [email protected] [email protected]
[email protected]
*YVO: [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
Step 5. Insert the %glo and the %tra dependent tiers
@Participants: YVO housewife1, NAT housewife2
@Languages: English (1), Spanish (2)
*YVO: [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected]?
*NAT: [email protected] [email protected] mí@2 [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
%glo: Yvonne for me not go to ask scones of these that now me are trying to
control a little-bit before of Christmas
Yvonne, don’t order these scones for me because now I am trying not
to put on weight before Christmas
*YVO: sí@2 [email protected] [email protected] está@2 [email protected] [email protected] [email protected]
[email protected]
%glo: yes Christmas already is round the corner woman
mind you, Christmas is already round the corner
*YVO: [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
[email protected] [email protected] [email protected] [email protected] [email protected] [email protected] [email protected]
Coding and analysing multilingual data
I already not make diet until for the less January or February and that
with luck
I am not going on a diet until at least January or February and even
then with a bit of luck
Step 6. Run the CHECK Programs again.
Once the transcription and tagging is completed, the next stage is to check
again the overall structure of the files, and the symbols and codes used in the
main and dependent tiers as declared in the CHILDES depfile and the LIDES
depadd file.
Step 7. Make changes in the depadd file.
Eventually, it may be necessary to add symbols to the LIDES depadd file for
your own special data-set of files. In this case the depadd file must be
changed using an ASCII editor and then you can run the CHECK programs
once again.
Step 8. Create a readme document to accompany the transcription file.
Using CHAT for language interaction data
The transcription and coding requirements of language interaction data differ
according to whether the language pairs under study are genetically related,
typologically similar or whether the mixed pairs are isolating, agglutinative,
inflective or incorporative languages. This section presents some suggestions as
to how to deal with specific transcription issues faced by researchers studying
language interaction.
Expanding the language tags
The language tag can be expanded, by using more numbers. You could, for
example, use the language tag not only to signal the languages used, but also to
assign a tag to mixed words. You could also use the language tag to code singleword calques, borrowings and so on. Aside from expanding the language tag for
language interaction types, and/or mixed words, you can also expand the language tag to a two (or more) digit system, where, for example, the first digit
denotes the language of the word, and the second digit word class, or vice versa.
Coding at morpheme level
For some research in language interaction data it is necessary to look not only at
words, but also at smaller morpheme units. There are two ways in CHAT to
dissect words into morphemes. The first is to code morphemes on the main tier.
Alternatively, you can use the %mor tier for more extensive morphological
coding. For more information on coding on the %mor tier we refer the reader to
MacWhinney (1995).
Coding turns
Turns are not basic units in CHAT, and without adaptations (like separately
coding for turns) the CLAN programs cannot make analyses pertaining to turns.
There are, however, various options for representing turns in a transcription file,
for example, by representing turns as a main tier.
Automated data analysis
The CLAN tools provide many possibilities for automatic analysis, although
they are not a substitute for the researcher’s efforts in analysing and interpreting
the data. They are especially helpful when large amounts of language data have
to be processed. Following we present a few of the most obvious applications of
the CLAN tools to multilingual data. There is extensive information on this in
the CHILDES manual. Three of the most useful analytical tools are FREQ,
FREQ frequency counts
The FREQ program makes frequency counts, which are useful for many different purposes. FREQ has properties which make this tool effective in handling
multilingual files for other types of counts as well, but we will start with
straightforward word counts.
COMBO pattern matching
COMBO is a CLAN tool which provides the user with ways of composing
complex search strings. The strings or patterns defined are used to find matches
of letters, words or groups of words in the data files specified. With the COMBO
tool the language tags on the main tier can be used, for example, for capturing
switches within utterances.
FLO output for other programs
There are sometimes advantages in looking at the data in their “pure” form,
particularly when a large amount of information and coding is added through
the dependent tiers. By applying the FLO tool all coding and other information
on the dependent tiers is left out. This output file can be made the input file for
the statistical package SPSS.
Coding and analysing multilingual data
Table A.1. Switched phrasal categories by language in oral and written texts,
in percentages
Switched phrasal categories:
Noun phrase:
Verb phrase:
Adjective phrase:
Prepositional phrase:
Adverbial phrase:
(Absolute number):
Oral texts
Written texts
Percentage of Total
Source: Moyer (2000)
A comparison of the Gibraltar oral corpus with the Gibraltar written
corpus (Moyer, 2000) illustrates how FREQ and COMBO, two of the
most useful CLAN programs, yield quantified information on the major
differences between oral and written code-switching styles.8 Prior classification and codification of the data following the LIDES recommendations
according to major structural and conversational units (that is, word, phrasal
constituents, utterances and turns) by language is needed for running the
CLAN programs.
The Gibraltar oral corpus is made up of sixteen recordings from eight different contexts in
Gibraltar. A total of 6,765 words have been transcribed. The written corpus is made up of sixteen
written constructions of oral speech used by two speakers. The size of this corpus is 5,049
Table A.2. Turns by language in oral and written texts, in percentages of turns
Language of turns:
Oral texts
Written texts
Percentage of Total
Equal number*
(Absolute number)
“Equal number” refers to turns that have the same number of words in both English and Spanish.
Source: Moyer (2000)
The FREQ program can tell us about: (a) the language that dominates in each
corpus; (b) the number and type of syntactic constituent in each language; (c)
the number of utterances by speaker (and language); and (d) the number of turns
by speaker (and language). Table A.1 shows the output of a FREQ program
showing in percentages the language of switched phrases in the oral and the
written corpora.
The COMBO program allows for more complex searches such as the combination of language switches within an utterance or turn of a given speaker. Table A.2
illustrates the results in percentages for the combination of language and turn in oral
and written code-switching styles.
Corpus-based studies and quantitative information about code-switching
patterns provide important empirical support for formal and interpretative
claims. Furthermore, the LIDES project offers data sets from a variety of
language pairs as well as the necessary tools for carrying out the analysis.
Practical information
Datasets currently available in the LIDES database
The LIDES database contains sample corpora with the following language combinations: Dutch–Turkish, Dutch–French, Alsatian German–French, Catalan–
Spanish, Catalan–Spanish–English, English–Spanish, English–German, English–
Punjabi, English–Greek Cypriot Dialect, English–Jamaican Creole.
These corpora vary widely in their size and their degree of development. For
more up to date and detailed information, please consult the LIPPS website
indicated below.
Coding and analysing multilingual data
Research contacts
Researchers interested in participating in the LIPPS group and the LIDES
database are advised to contact the steering committee. A list of email addresses
is provided below.
Penelope Gardner-Chloros: [email protected]
Roeland van Hout: [email protected]
Melissa G. Moyer: [email protected]
Mark Sebba: [email protected]
More information can be found on the LIPPS homepage on the World
Wide Web. The URL for the LIPPS homepage is: www.ling.lancs.
How to obtain the CLAN programs
You can download the CLAN tools together with a CHAT depfile from the
CHILDES Internet pages: http://childes.psy.cmu.edu
Glossary of terms
(This glossary is adapted from the LIDES Coding Manual, International
Journal of Bilingualism (2000) 4(2):131–270)
A concept introduced by Muysken (1995) to characterize
code-switching from a psycholinguistic perspective within a
sentence or between sentences. It involves the use of
stretches of language from different systems which do not
necessarily make up a syntactic constituent. Alternation
contrasts with insertion and congruent lexicalization. In
insertion, the grammar of a single language predominates,
incorporating a word or larger syntactic constituent from
another language. In congruent lexicalization, the syntax of
the two languages in the sentence is the same, thus allowing
the lexical items from either language to be used.
In language interaction research lexical borrowing
typically refers to the linguistic forms being taken over
by one language or language variety from another. A
crucial methodological issue is the distinction between
borrowing and code-switching. Difficulties involved in
this distinction include estimating, and deciding the
relevance of, the degree of morphological, phonological
and syntactic integration of the borrowed item.
CLAN programs
A CHAT data file may contain information which changes
within the file (for example, a change in the activity being
carried out by the participants). Changeable headers may be
placed within the body of the transcription file and they are
preceded by the symbol @ (for example, @Activities: PAT
tries to get CHI to put on her coat). Changeable headers
should be distinguished from constant and obligatory file
headers which are preceded by the same symbol @.
Obligatory file headers are essential if a researcher wants
to use the CLAN automatic analysis programs (see 4.1.1).
The transcription system created by MacWhinney (1991,
1995) for the CHILDES project. This transcription scheme
has been adapted and elaborated further by LIDES to deal
more precisely with the theoretical and practical problems
raised by language interaction data.
The analytical tools developed by MacWhinney (1991,
1995) for the CHILDES project, which can be used by
researchers to carry out automated analyses of their data
This is a general term which refers to the alternate use of
two or more languages or language varieties by bilinguals
for communicative purposes. Code-switching embraces
various types of bilingual behavior such as switching
within and between utterances, turns and sentences. A
theoretically neutral and less confusing term adopted for
code-switching is language interaction.
The main use of this CLAN program is to search
transcription files for specific expressions, words or
strings. The COMBO search command is constructed
with a set of Boolean operators (AND, OR, NOT, etc.)
which are used to define the parameters of the search
string, which may be some aspect of the coding (that is
language tags) or words or utterances from the text data
file. Searches may be carried out on main or dependent
tiers. An example of a COMBO search command to find
all lexical items which cannot be assigned to a given
language and which are coded on the main tier as @0 is:
combo +s *\ @0 ^ * ^ * gibraltar.doc OJO
A CHAT data file may contain non-obligatory constant
headers at the beginning of the file in order to specify
information relevant to the conversation text transcribed.
Coding and analysing multilingual data
depadd file
dependent tier
file headers
main tier
Such information may include gender or age of
participants, level of education, social economic status,
information about coders and transcribers, and languages
spoken by each of the participants. As with the case of
obligatory and changeable headers, constant headers are
preceded by the symbol @ such as: @Gender of
participants: CHI female, PAT male
The file where a researcher using CHAT incorporates the
list of their newly created symbols or coding schemes used
in their transcription. Adding symbols to the depadd file is
easy and it gives flexibility to the researcher to develop
his/her own system of coding. The depadd file, like the
depfile, is used by the CHECK program to verify the
syntax and structure of data files.
The CHAT transcription scheme makes use of dependent
tiers to include information or additional coding (that is
syntactic, morphological, speech acts, general comments
and so on) which refer to the preceding main tier. It is
possible to have any number of dependent tiers at the
same time which make reference to a single main tier. All
dependent tiers should be preceded by the symbol %
followed by a three letter code among those already
defined in the depfile or a newly created code which
should be added to the depadd file. An example of a
dependent tier code for a gloss of the main tier utterance
is: %glo.
One of the main components of a CHAT transcript are the
file headers. These are lines of text preceded by the symbol
@ (for example, @Participants). Headers can be obligatory
(such as @Begin, or @End) or optional (such as
@Participants, or @Coder). Different types of headers
can be inserted in different places in a CHAT transcription.
A morpheme by morpheme translation of one or more
utterances on the main tier. This gloss is strongly
recommended for researchers contributing their data to
the database and it should be incorporated on a specific
dependent tier (%glo).
This is the line in a CHAT transcription which includes a
person’s speech transcribed in written form. It is preceded
by an asterisk, the speaker’s initials or other three-letter code
(for example, *PAM:). Each turn of the conversation may
be divided into several entries or the speech may be
transcribed in a single entry
May also be referred to as a meta-symbol. A meta-character
is a symbol used in coding a transcription in CHAT (for
example, %, *). Meta-symbols or characters are used
for various types of coding in the headers, the main tier
and dependent tiers. Any non-letter character is a meta
readme file
Provides important background information about the
files belonging to a data-set. This information is crucial
for other researchers working with this data. The original
researcher may include any comment which is important
for understanding the particular data set or the particular
way it has been transcribed and coded. The readme file is a
separate document or file which is included with all data
contributed to the LIDES database.
This acronym stands for Standard Generalized Mark-up
Language. It is a meta-language which is used as a standard
way of marking up a text (both spoken and written), and is
independent of any word processor or computer system.
XML (Extensible
This language is a flexible text format derived from
Markup Language) SGML (see above). XML is a metalanguage written in
SGML which allows a user to design customized markup
languages. XML now provides the framework for the
CHILDES and TALKBANK databases. See http://www.
w3.org/XML/ for more information.
Aaho, T. 1999. Code-switching in the Greek Cypriot community in London.
Unpublished MA thesis, University of Helsinki, Dept. of English.
Adams, J. N. 2003. Bilingualism and the Latin Language. Cambridge: Cambridge
University Press.
Adams, J. N., Janse, M. & Swain, S. (eds) 2002. Bilingualism in Ancient Society:
language contact and the written text. Oxford: Oxford University Press.
Adone, D. 1994. The Acquisition of Mauritian Creole. Amsterdam: John Benjamins.
Agnihotri, R. K. 1987. Crisis of Identity: The Sikhs in England. New Delhi: Bahri.
Aikhenvald, A. Y. 2002. Language Contact in Amazonia. Oxford: Oxford University
Aitchison, J. 1994. Words in the Mind, 2nd ed. Oxford: Blackwell.
2000. Language Change: Progress or Decay? 3rd ed. Cambridge: Cambridge
University Press.
Aitsiselmi, F. 2003. L’alternance codique sur Beur FM. Talk given at French Linguistics
Day, London Metropolitan University, 15 March 2003.
Albert, M. L. & Obler, L. K. 1978. The Bilingual Brain: neuropsychological and neurolinguistic aspects of bilingualism. New York: Academic Press.
Alfonzetti, G. 1998. The conversational dimension in code-switching between Italian
and dialect in Sicily. In P. Auer 1998b (ed.), 180–214.
Alvarez-Cáccamo, C. 1998. From ‘switching code’ to ‘code-switching’: towards a
reconceptualization of communicative codes. In P. Auer (ed.) 1998c, 29–50.
Anderson, B. 1991. Imagined Communities: Reflections on the Origins and Spread of
Nationalism. London and New York: Verso.
Angermeyer, P. S. 2005. Spelling bilingualism: script choice in Russian American
classified ads and signage. Language in Society 34(4): 493–531.
Ann, J. 2001. Bilingualism and language contact. In C. Lucas (ed.), The Sociolinguistics
of Sign Languages. Cambridge: Cambridge University Press.
Anthias, F. 1992. Ethnicity, Class, Gender and Migration. Aldershot: Avebury.
Appel, R. & Muysken. P. 1987. Language Contact and Bilingualism. London: Edward
Arnberg, L. N. & Arnberg, P. W. 1992. Language awareness and language separation in
the young bilingual child. In R. J. Harris (ed.), Cognitive Processes in Bilinguals.
Oxford: Elsevier, 475–500.
Aslanov, C. 2000. Interpreting the language-mixing in terms of codeswitching: The
case of the Franco-Italian interface in the Middle Ages. Journal of Pragmatics 32,
Atkinson, J. M. & Heritage, J. 1984. Structures of Social Action: studies in conversation
analysis. Cambridge: Cambridge University Press.
Auer, P. 1984. Bilingual Conversation. Amsterdam/Philadelphia: John Benjamins.
1995. The pragmatics of code-switching: a sequential approach. In Milroy &
Muysken, 1995, 115–135.
1998a. ‘Bilingual Conversation’ revisited. In Auer, P. 1998c, 1–25.
1998b (ed.). Code-switching in Conversation: Language, Interaction and Identity.
London: Routledge.
1999. From codeswitching via language mixing to fused lects: toward a dynamic
typology of bilingual speech. International Journal of Bilingualism, 3(4), 309–332.
2000. Why should we and how can we determine the ‘base language’ of a bilingual
conversation? Estudios de Sociolinguística, 1(1), 129–144.
2005. A postscript: code-switching and social identity. Journal of Pragmatics 37,
Azuma, S. 1996. Speech production units among bilinguals. Journal of Psycholinguistic
Research 25(3), 397–416.
1997. Lexical categories and code-switching: a study of Japanese–English codeswitching in Japan. Journal of the Association of Teachers of Japanese, 31(2), 1–24.
1998. Meaning and form in code-switching. In R. Jacobson, 1998a, 109–125.
Backus, A. 1992. Patterns of Language Mixing: a study of Turkish–Dutch bilingualism.
Wiesbaden: Harrassowitz.
1996. Two in One: bilingual speech of Turkish immigrants in the Netherlands. Tilburg:
Tilburg University Press.
1999. The intergenerational codeswitching continuum in an immigrant community. In
Extra & Verhoeven, 1999, 261–279.
2003. Units in codeswitching: evidence for multimorphemic elements in the lexicon.
Linguistics 41(1), 83–132.
2004. Convergence as a mechanism of language change. Bilingualism, Language and
Cognition 7(2), 179–181.
2005. Codeswitching and language change: one thing leads to another? International
Journal of Bilingualism 9(3/4), 307–341.
Backus, A. & van Hout, R. 1995. Distribution of code-switches in bilingual conversations. In Proceedings of the Summer School on Codeswitching and Language
Contact. Leeuwaarden: Fryske Akademy, pp. 16–28.
Badequano-López, P. & Kattan, S. 2007. Growing up in a multilingual community:
insights from language socialization. In P. Auer & Li Wei (eds), Handbook of
Multilingualism and Multilingual Communication. Berlin: Mouton de Gruyter,
Bader, Y. & Minnis, D. D. 2000. Morphological and syntactic code-switching in the
speech of an Arabic–English bilingual child. Multilingua 19(4), 383–403.
Baetens Beardsmore, H. 1986. Bilingualism: basic principles, 2nd ed. Clevedon:
Multilingual Matters.
Baetens Beardsmore, H. & Anselmi, G. 1991. Code-switching in a heterogeneous,
unstable, multilingual speech community. In Papers for the Symposium on Codeswitching in Bilingual Studies: theory, significance and perspectives, March, 1991.
Barcelona: European Science Foundation, 405–436.
Baker, C. & Prys Jones, S. 1998. Encyclopedia of Bilingualism and Bilingual Education.
Clevedon: Multilingual Matters.
Bakhtin, M. 1986. The problem of speech genres. In C. Emerson & M. Homquist
(eds), Speech Genres and Other Late Essays. Austin: University of Texas Press,
Bakker, P. 1997. A Language of Our Own: the genesis of Michif, the mixed Cree–French
language of the Canadian Metis. Oxford: Oxford University Press.
Bakker, P. & Mous, M. 1994 (eds). Mixed Languages: fifteen case studies in language
intertwining. Studies in Language and Language Use, 13. Amsterdam: Institute for
Functional Research into Language and Language Use.
Barrett, R. 1998. Markedness and styleswitching in performances by African American
Drag Queens. In Myers-Scotton, 1998, 139–167.
Barron-Hauwert, S. 2004. Language Strategies for Bilingual Families: the one-parentone-language approach. Clevedon: Multilingual Matters.
Bauer, E. B., Hall, K. J. & Kruth, K. 2002. The pragmatic role of codeswitching in play
contexts. International Journal of Bilingualism 6(1), 53–74.
Baugh, A. C. 1951. A History of the English Language. London: Routledge.
Belazi, H. M., Ruben, J. R. & Toribio, A. J. 1994. Code-switching and X-bar theory: the
functional head constraint. Linguistic Enquiry 25, 221–237.
Bell, A. 1984. Language style as audience design. Language in Society 13, 145–204.
1997. Language style as audience design. In N. Coupland & A. Jaworski (eds)1997,
Sociolinguistics: a reader. New York: St Martin’s Press, 240–250.
2001. Back in style: reworking audience design. In Eckert & Rickford, 2001,
Benoit, E., Randolph, D., Dunlap, E. & Johnson, B. 2003. Code-switching and inverse
imitation among marijuana-using crack sellers. British Journal of Criminology 43,
Benson, E. J. 2001. The neglected early history of codeswitching research in the United
States. Language and Communication 21(1), 23–36.
Bentahila, A. 1983. Language Attitudes Among Arabic–French Bilinguals in Morocco.
Clevedon: Multilingual Matters.
Bentahila, A. & Davies, E. E. 1983. The syntax of Arabic–French code-switching.
Lingua 59, 301–330.
1991. Constraints on code-switching: a look beyond grammar. Papers for the
Symposium on Code-switching in Bilingual Studies: theory, significance and perspectives. March, 1991. Barcelona: European Science Foundation, 369–405.
1998. Codeswitching: an unequal partnership? In Jacobson, 1998a, 25–51.
2002. Language mixing in rai music: localization or globalization? Language and
Communication 22, 187–207.
Berent, G. P. 2004. Sign language–spoken language bilingualism: code-mixing and
mode-mixing by ASL–English bilinguals. In Bhatia & Ritchie 2004, 312–336.
Besnier, N. 2003. Crossing genders, mixing languages: the linguistic construction of
transgenderism in Tonga. In J. Holmes & M. Meyerhoff (eds), The Handbook of
Language and Gender. Oxford: Blackwell, 279–301.
Bhatia, T. K. 1992. Discourse functions and pragmatics of mixing: advertising across
cultures. World Englishes 11 (2/3), 195–215.
Bhatia, T. K. & Ritchie, W. C. 1996. Bilingual language mixing, Universal Grammar,
and second language acquisition. In W. C. Ritchie & T. K. Bhatia (eds), The
Handbook of Second Language Acquisition. San Diego/London: Academic Press,
2004 (eds). Handbook of Bilingualism. Oxford: Blackwell.
Bialystok, E. 1991 (ed.). Language Processing in Bilingual Children. Cambridge:
Cambridge University Press.
Biber, D. & Finegan, E. 1993 (eds). Sociolinguistic Perspectives on Register. New York:
Oxford University Press.
Biber, D., Johansson, S., Leech, G., Conrad, S. & Finegan, E. 1999. Longman Grammar
of Spoken and Written English. Harlow: Longman.
Bickerton, D. 1981. Roots of Language. Ann Arbor: Karoma.
Bister-Broosen, H. 2002. Alsace. In J. Treffers-Daller & R. Willemyns (eds), Language
Contact at the Romance–Germanic Language Border. Clevedon: Multilingual
Matters, 98–111.
Blom, J. P. & Gumperz, J. J. 1972. Social meaning in linguistic structures: codeswitching in Norway. In Gumperz & Hymes, 1972, 407–434. Reprinted in Li
Wei, 2000, 111–136.
Boeschoten, H. 1994. L2 influence on L1 development: the case of Turkish children in
Germany. In G. Extra & L. Verhoeven 1994 (eds), The Cross-Linguistic Study of
Bilingual Development. Amsterdam: North-Holland, 253–264.
1998. Codeswitching, codemixing, and code alternation: what a difference. In
Jacobson, 1998a, 15–24.
Boeschoten, H. E. & Verhoeven, L. T. 1987. Language mixing in children’s speech:
Dutch language use in Turkish discourse. Language Learning 37(2), 191–217.
Bokamba, E. 1988. Code-mixing, language variation and linguistic theory: evidence
from Bantu languages. Lingua 76, 21–62.
Bolonyai, A. 1998. In-between languages: language shift/maintenance in childhood
bilingualism. International Journal of Bilingualism 2(1), 21–43.
Bosch, L. & Sebastián-Gallés, N. 2001. Early language differentiation in bilingual
infants. In Cenoz & Genesee, 2001, 71–94.
Botero, C., Bullock, B., Davis, K. & Toribio, J. A. 2004. Perseverative phonetic effects in
bilingual code-switching. Paper presented at the 34th Linguistic Symposium on
Romance Languages, University of Utah, March.
Bourdieu, P. 1997. Language and Symbolic Power. Oxford: Polity Press.
Brazil, D. 1995. A Grammar of Speech. Oxford: Oxford University Press.
Breitborde, L. B. 1982. Levels of analysis in sociolinguistic explanation: bilingual
codeswitching, social relations and domain theory. International Journal of the
Sociology of Language 39, 4–45.
Broca, P. 1861. Remarques sur le siège de la faculté du langage articulé, suivies d’une
observation d’aphémie (perte de la parole). Bulletin de la Société Anatomique 6,
Broeder, P. & Extra, G. 1995. Minderheldsgroepen en Minderheidstalen. Den Haag:
Günther Narr Verlag.
Broersma, M. & de Bot, K. 2006. Triggered codeswitching: a corpus-based evaluation of
the original triggering hypothesis and a new alternative. Bilingualism, Language
and Cognition 9(1), 1–13.
Brown, P. 1994. Gender, politeness, and confrontation in Tenejapa. In D. Tannen (ed.),
Gender and Conversational Interaction. New York: Oxford University Press,
Brown, P. & Levinson, S. 1987. Politeness: some universals in language use.
Cambridge/New York: Cambridge University Press.
Brown, R. & Gilman, A. 1960. The pronouns of power and solidarity. In P. Giglioli (ed.),
Language and Social Context: Selected Readings, Harmondsworth: Penguin, 252–282.
Brulard, I. & Carr, P. 2003. French–English bilingual acquisition of phonology: one
production system or two? International Journal of Bilingualism 7(2), 177–202.
Bullock, B. & Toribio, A. J. 2004. Introduction: convergence as an emergent property in
bilingual speech. Bilingualism, Language and Cognition 7(2), 91–93.
Burt, S. M. 2002. Maxim confluence. Journal of Pragmatics 34, 993–1001.
Bynon, T. 1977. Historical Linguistics. Cambridge: Cambridge University Press.
Cameron, D. 1992. Feminism and Linguistic Theory. London: Macmillan/New York:
St Martin’s.
1995. Verbal Hygiene. London/New York: Routledge.
1997. Demythologizing sociolinguistics. In Coupland & Jaworski, 1997, 55–68.
Camilleri, A. 1996. Language values and identities: code-switching in secondary classrooms in Malta. Linguistics and Education 8(1), 85–103.
Canut, C. & Caubet, D. 2001. Comment les langues se mélangent: codeswitching en
francophonie. Paris: L’Harmattan.
Cashman, H. 2005. Identities at play: language preference and group membership in
bilingual talk in interaction. Journal of Pragmatics 37, 301–315.
Caubet, D. 2001. Comment apprehender le code-switching? In Canut & Caubet, 2001,
Cenoz, J. & Genesee, F. 2001 (eds). Trends in Bilingual Acquisition. Amsterdam:
Cenoz, J., Hufeisen, B. & Jessner, U. 2001 (eds). Cross-linguistic Influences in
Trilingualism and Third Language Acquisition. Clevedon: Multilingual Matters.
Cenoz, J. & Jessner, U. 2000 (eds). English in Europe: The Acquisition of a Third
Language. Clevedon: Multilingual Matters.
Chambers, J. K. 2003. Sociolinguistic Theory, 2nd ed. Oxford: Blackwell.
Chambers, J. K., & Trudgill, P. 1999. Dialectology, 2nd ed. Cambridge: Cambridge
University Press.
Chan, B. 1999. Aspects of the syntax, production and pragmatics of code-switching with
special reference to Cantonese–English. Unpublished PhD thesis, Department of
Linguistics, University College London.
Chana, U. & Romaine, S. 1984. Evaluative reactions to Panjabi–English code-switching.
Journal of Multilingual and Multicultural Development 5, 447–453.
Charles, R. 1995. Code-switching in the Punjabi–English community in Britain: a mixed
code. Unpublished MA dissertation, Department of Applied Linguistics, Birkbeck
College, London.
Cheng, Karen Kow Yip 2003. Code-switching for a purpose: focus on pre-school
Malaysian children. Multilingua 22(1), 59–77.
Cheshire, J. 2004. Old and new cultures in contact: approaches to the study of syntactic
variation and change. Paper given at Sociolinguistics Symposium 15, University of
Newcastle, 1–4 April, 2004.
Cheshire, J. & Gardner-Chloros, P. 1997. Communicating gender in two languages. In
H. Kothoff & R. Wodak (eds), Communicating Gender in Context. Amsterdam:
Benjamins, 249–283.
1998. Code-switching and the sociolinguistic gender pattern. International Journal of
the Sociology of Language 129, 5–34.
Cheshire, J. & Moser, L. 1994. English as a cultural symbol: the case of advertisements in
French-speaking Switzerland. Journal of Multilingual and Multicultural
Development 15, 451–469.
Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
1981. Lectures on Government and Binding. Dordrecht: Foris.
1986. Knowledge of Language: its nature, origin and use. New York: Praeger.
1995. The Minimalist Program. Cambridge, MA: MIT Press.
Christodoulou-Pipis, I. 1991. Greek Outside Greece III. Research Findings: language use by
Greek-Cypriots in Britain. Nicosia: Diaspora Books, Greek Language Research Group.
Cicero, Marcus Tullius 1999. Letters to Atticus. Edited and translated by
D. R. Shackleton Bailey. Cambridge, MA/London: Harvard University Press.
Clark, H. H. & Clark, E. V. 1977. Psychology and Language: an introduction to psycholinguistics. New York: Harcourt Brace Jovanovich.
Cleghorn, A. 1992. Primary level science in Kenya: constructing meaning through
English and indigenous languages. International Journal of Qualitative Studies in
Education 5(4), 311–323.
Clyne, M. 1967. Transference and Triggering: observations on the language assimilation
of postwar German-speaking migrants in Australia. The Hague: Martinus Nijhoff.
1972. Perspectives on Language Contact: a study of German in Australia. Melbourne:
1987. Constraints on code switching: how universal are they? Linguistics 25,
1991a. Structural and typological aspects of community languages. In Clyne, 1991b,
1991b. Community Languages: the Australian experience. Cambridge: Cambridge
University Press.
2003. Dynamics of Language Contact: English and immigrant languages.
Cambridge: Cambridge University Press.
Coates J. 1993. Women, Men and Language. London: Longman.
1996. Women’s Talk: conversations between women friends. Oxford: Blackwell.
Colina, S. & MacSwan, J. 2005. Language Mixing at the Interface: How Does
Phonology Affect the Syntax of Codeswitching? Paper presented at the 5th
International Symposium on Bilingualism (ISB5), Barcelona, Spain.
Comeau, L., Genesee, F. & Lapaquette, L. 2003. The modelling hypothesis and child
bilingual codemixing. International Journal of Bilingualism 7(2), 113–126.
Coulmas, F. 1986. Reported speech: some general issues. In F. Coulmas (ed.), Direct and
Indirect Speech. Berlin: Mouton de Gruyter, 59–77.
1997 (ed.). The Handbook of Sociolinguistics. Oxford: Blackwell.
Coupland, N. 1985. Hark hark the lark: social motivations for phonological styleshifting. Language and Communication 5, 153–171.
2001. Language, situation and the relational self: Theorizing dialect-style in sociolinguistics. In P. Eckert & J.R. Rickford (eds), 185–211.
2007. Style: Language variation and identity. Cambridge: Cambridge University
Coupland, N. & Jaworski, A. 1997. Sociolinguistics: a reader and coursebook.
Basingstoke/London: Macmillan.
Cowie, F. 1999. What’s within? Nativism Reconsidered. Oxford: Oxford University
Cromdal, J. 2001. Overlap in bilingual play: some implications of code-switching for
overlap resolution. Research on Language and Social Interaction 34(4), 421–451.
Crowley, T. 1990. Beach-la-Mar to Bislama: emergence of a national language in
Vanuatu. Oxford: Oxford University Press.
Crystal, D. 2000. Language Death. Cambridge: Cambridge University Press.
Cummins, J. 2003. Bilingual education: basic principles. In Dewaele, Housen & Li Wei,
2003, 56–67.
Dabène, L. & Moore, D. 1995. Bilingual speech of migrant people. In Milroy &
Muysken, 1995, 17–44.
Davis, J. 1989. Distinguishing language contact phenomena in ASL interpretation. In
C. Lucas (ed.), The Sociolinguistics of the Deaf Community. New York: Academic
Press/Harcourt Brace Jovanovich, 85–103.
De Bose, C. 1992. Codeswitching: black English and Standard English in the
African–American linguistic repertoire. Journal of Multilingual and Multicultural
Development 13, 157–167.
De Bot, K. 1992/2000. A bilingual production model: Levelt’s ‘speaking’ model adapted. Applied Linguistics 13, 1–24. Reproduced in Li Wei, 2000, 420–442.
2004. The multilingual lexicon: modelling selection and control. International
Journal of Multilingualism, 1(1), 17–32.
De Bot, K. & Schreuder, R. 1993. Word production and the bilingual lexicon. In
Schreuder & Weltens, 1993, 191–214.
De Groot, A. M. B. & Kroll, X. 1997. Tutorials in Bilingualism: psycholinguistic
perspectives. Mahwah, NJ: Lawrence Erlbaum.
De Houwer, A. 1990. The Acquisition of Two Languages from Birth: a case study.
Cambridge: Cambridge University Press.
1995. Bilingual language acquisition. In Fletcher & MacWhinney, 1995, 219–250.
2009. Bilingual First Language Acquisition. Clevedon: Multilingual Matters.
Dell, G. 1986. A spreading activation theory of retrieval in sentence production.
Psychological Review 93, 283–321.
De Mejía, A.-M. 2002. Power, Prestige and Bilingualism: international perspectives on
elite bilingual education. Clevedon: Multilingual Matters.
Denison, N. 1972. Some observations on language variety and plurilingualism. In
J. B. Pride & J. Holmes (eds), Sociolinguistics. Harmondsworth: Penguin, 65–77.
1984. Language acquisition in a plurilingual environment. Grazer linguistische
Studien 21, 35–64.
1986. Sociolinguistics, linguistic description, language change and language acquisition. In J. Fishman (ed.), The Fergusonian Impact. Vol. 1: From phonology to
society. Berlin: Mouton de Gruyter, 83–98.
1987. Sauris, a typical ‘linguistic island’ in the Carnian Alps. In Isole linguistiche e
culturali, Proceedings of the 24th AIMAV Colloquy, University of Udine, 13–16
May, 65–75.
Deuchar, M. 2006. Welsh–English code-switching and the Matrix Language frame
model. Lingua 116(11), 1986–2011.
Deuchar, M. & Quay, S. 2000. Bilingual Acquisition: theoretical implications of a casestudy. Oxford: Oxford University Press.
De Villiers P. A. & De Villiers J. G. 1979. Early Language. Fontana Paperbacks.
Dewaele, J.-M. 2001. Activation or inhibition? The interaction of L1, L2 and L3 on the
language mode continuum. In U. Jessner, B. Hufeisen & J. Cenoz (eds), Crosslinguistic Influence in Third Language Acquisition: psycholinguistic perspectives.
Clevedon: Multilingual Matters, 69–89.
2003. Code-switching and emotion talk among multilinguals. Paper presented at the
4th International Symposium on Bilingualism, Phoenix, Arizona.
2004. The emotional force of swearwords and taboo words in the speech of multilinguals. Journal of Multilingual and Multicultural Development 25(2/3), 204–222.
2005. Investigating the psychological and emotional dimensions in Instructed
Language Learning: obstacles and possibilities. Modern Language Journal
89(iii), 367–380.
Dewaele, J.-M. & Edwards, M. 2000. Conversations trilingues (arabe, français, anglais)
entre mère et fille: un saute-Mouton Mouton de Gruyter référentiel? In C. Charner
(ed.), Communications référentielles et processus référentiels. Langues, cultures,
interprétations. Série Le fil du discours. Publications de Praxiling –Université PaulValéry, Montpellier, 193–214.
Dewaele, J.-M., Housen, A. & Li Wei 2003 (eds). Bilingualism: beyond basic principles.
Festschrift in honour of Hugo Baetens Beardsmore. Clevedon: Multilingual
Diaz, R. M. 1983. Thought in two languages: the impact of bilingualism on cognitive
development. In E. W. Gordon (ed.), Review of Research in Education. Vol. X.
Washington, DC: AERA.
Dijkstra, T. & van Hell, J. 2003. Testing the Language Mode Hypothesis using trilinguals. International Journal of Bilingual Education and Bilingualism, 6(1), 2–16.
D’Introno, F. 1996. Spanish–English code-switching: conditions on movement. In Roca
& Jensen, 1996, 187–201.
Di Sciullo, A. M., Muysken, P. & Singh, R. 1986. Government and code-mixing. Journal
of Linguistics 22, 1–24.
Dolitsky, M. 2000. Codeswitching in a child’s monologues. Journal of Pragmatics 32,
Döpke, S. 2000 (ed.). Cross-linguistic Structures in Simultaneous Bilingualism.
Amsterdam/Philadelphia: Benjamins.
Dorian, N. 1981. The Life Cycle of a Scottish Gaelic Dialect. Philadelphia: University of
Pennsylvania Press.
Doron, E. 1983. On a formal model of code-switching. Texas Linguistic Forum 22, 35–59.
Dressler, W. U. & Wodak-Leodolter, R. 1977. Introduction to: Language death.
International Journal of the Sociology of Language 12, 5–12.
Duranti, A. 1988. Intentions, language and social action in a Samoan context. Journal of
Pragmatics 12, 13–33.
Dussias, P. E. 2001. Psycholinguistic complexity in codeswitching. International
Journal of Bilingualism 5(1), 87–100.
Eastman, C. M. 1992 (ed.). Codeswitching. Clevedon: Multilingual Matters.
Eckert, P. 1989. The whole woman: sex and gender differences in variation. Language
Variation and Change 1(3), 245–268.
1996. (ay) goes to the city: exploring the expressive use of variation. In G. R. Guy,
C. Feagin, D. Schiffrin & J. Baugh (eds), Towards a Social Science of Language.
Volume 1. Variation and change in language and society. Amsterdam: Benjamins,
2000. Linguistic Variation as Social Practice. Oxford/Malden, MA: Blackwell.
Eckert, P. & Rickford, J. R. 2001 (eds). Style and Sociolinguistic Variation. Cambridge:
Cambridge University Press.
Edwards, J. 1995. Principles and alternative systems in the transcription, coding and
mark-up of spoken discourse. In Leech, Myers & Thomas, 1995, 19–34.
Edwards, M. & Gardner-Chloros, P. 2007. Compound verbs in code-switching: bilinguals making do? International Journal of Bilingualism, 11(1), 73–91.
Eilers, R. E., Gavin, W. J. & Oller, D. K. 1982. Cross-linguistic perception in infancy:
early effects of linguistic experience. Journal of Child Language 9, 289–302.
Eisikovits E. 1991. Variation in subject–verb agreement in Inner Sydney English. In
J. Cheshire (ed.), English Around the World: sociolinguistic perspectives.
Cambridge: Cambridge University Press, 227–234.
Eliasson, S. 1989. English–Maori Language contact: code-switching and the free morpheme constraint. Reports from Uppsala University Department of Linguistics
(RUUL) 18, 1–28.
1991. Models and constraints in code-switching theory. Papers from the Workshop on
Constraints, Conditions and Models. Strasbourg: European Science Foundation,
Eppler, E. 1999. Word order in German–English mixed discourse. UCL Working Papers
in Linguistics 11, 285–308.
Epstein, S. P., Groat, E. M., Kawashima, R. & Kitahara, H. 1998. A Derivational
Approach to Syntactic Relations. Oxford: Oxford University Press.
Ervin-Tripp, S. 2001. Variety, style-shifting, and ideology. In Eckert & Rickford, 2001,
Ervinn-Tripp, S. & Reyes, I. 2005. Child codeswitching and adult content contrasts.
International Journal of Bilingualism, Special Issue on Cognitive Approaches to
Bilingualism 9(1), 85–103.
Extra, G. & Verhoeven, L. 1999. Bilingualism and Migration. Berlin: Mouton de Gruyter.
Fabbro, F. 1999. The Neurolinguistics of Bilingualism: an introduction. Hove:
Psychology Press.
Fabbro, F., Skrap, M. & Aglioti, S. 2000. Pathological switching between languages after
frontal lesions in a bilingual patient. Journal of Neurology, Neurosurgery and
Psychiatry 68, 650–652.
Fano, R. M. 1950. The information theory point of view in speech communication.
Journal of the Acoustical Society of America 22, 691–696.
Fantini, A. E. 1985. Language Acquisition of a Bilingual Child: a sociolinguistic perspective (to age ten). Providence, RI: College Hill.
Farrar, K. & Jones, M. C. 2002. Introduction. In M. C. Jones & E. Esch (eds), Language
Change: the interplay of internal, external and extra-linguistic factors. Berlin/NY:
Mouton de Gruyter, 1–19.
Ferguson, C. 1959. Diglossia. Word 15, 325–337. Reprinted in Li Wei, 2000, 65–80.
Finlayson. R. & Slabbert, S. 1997. ‘I’ll meet you halfway with language’: codeswitching within a South African urban context. In M. Pütz (ed.), Language
Choices: conditions, constraints, and consequences. Amsterdam/Philadelphia:
John Benjamins, 381–423.
Finnis, K. 2006. Language use and socio-pragmatic meaning: code-switching amongst
British-born Greek-Cypriots. Unpublished PhD thesis, Birkbeck, University of
Fishman, J. 1965. Who speaks what language to whom and when? La linguistique 2,
67–87. Reprinted in Li Wei, 2000, 89–06.
1967. Bilingualism with and without diglossia, diglossia with and without bilingualism. Journal of Social Issues 23, 29–38.
1968. Language Problems of Developing Nations. New York: Wiley.
1972. Domains and the relationship between micro and macro sociolinguistics. In
Gumperz & Hymes, 1972, 435–453.
Fletcher, P. & MacWhinney, B. 1995. A Handbook of Child Language. Oxford: Blackwell.
Florence, S. 2003. A Guide to Reading and Writing Japanese: a comprehensive guide to
the Japanese writing system. Boston: Tuttle.
Foley, J. A. 1998. Code-switching and learning among young children in Singapore.
International Journal of the Sociology of Language 130, 129–150.
Franceschini, R. 1998. Code-switching and the notion of code in linguistics: proposals
for a dual focus model. In Auer, 1998c, 51–74.
Fries, C. C. & Pike, K. L. 1949. Coexistent phonemic systems. Language 25(1), 29–50.
Gafaranga, J. 2005. Demythologizing language alternation studies: conversational structure vs. social structure in bilingual interaction. Journal of Pragmatics 37, 281–300.
Gafaranga, J. & Torras, M.-C. 2001. Do speakers speak a language? Language versus
medium in the study of bilingual conversation. International Journal of
Bilingualism 5(2), 195–219.
2002. Interactional otherness: towards a redefinition of code-switching. International
Journal of Bilingualism 6(1), 1–22.
Gal, S. 1979. Language Shift: social determinants of linguistic change in bilingual
Austria. New York: Academic Press.
1988. The political economy of code choice. In M. Heller (ed.) Codeswitching: anthropological and sociolinguistic perspectives. Berlin: Mouton de Gruyter, 245–264.
Gardner-Chloros, P. 1987. Code-switching in relation to language contact and convergence. In G. Lüdi (ed.), Devenir bilingue-parler bilingue; actes du 2ème colloque
sur le bilinguisme, 20–22 septembre 1984. Tübingen: Niemeyer, 99–115.
1991. Language Selection and Switching in Strasbourg. Oxford: Oxford University Press.
1992. The sociolinguistics of the Greek-Cypriot Community in London. In
M. Karyolemou (ed.), Plurilinguismes: sociolinguistique du grec et de la Grèce.
No.4, juin. Paris: CERPL, 112–136.
1995. Code-switching in community, regional and national repertoires: the myth of the
discreteness of linguistic systems. In Milroy & Muysken, 1995, 68–90.
1997. Code-switching: language selection in three Strasbourg department stores. In
Coupland & Jaworski, 1997, 361–376.
1998. Code-switching and the sociolinguistic gender pattern (with J. Cheshire).
International Journal of the Sociology of Language. Special edition on Women’s
Languages in Various Parts of the World. 129, 5–34.
2000. The tortoise and the hare: distinguishing processes and end-products in language contact. Bilingualism, Language and Cognition 3(2), 112–114.
2004. Code-switching and Literacy: the case of a Brussels French/Dutch illiterate
bilingual. Sociolinguistics Symposium 15, University of Newcastle, 1–4 April.
Gardner-Chloros, P. & Charles, R. 2007. Subiko welcome (Welcome to everyone): Hindi/
English code-switching in the British-Asian media. BISAL (Birkbeck Studies in
Applied Linguistics) 2, 89–122. www.bisal.bbk.ac.uk/publications/volume2/pdf/
Gardner-Chloros, P., Charles, R. & Cheshire, J. 2000. Parallel patterns? A comparison of
monolingual speech and bilingual codeswitching discourse. Journal of Pragmatics.
Special Issue on Code-switching. 32, 1305–1341.
Gardner-Chloros, P. & Edwards, M. 2004. Assumptions behind grammatical approaches
to code-switching: when the blueprint is a red herring. Transactions of the
Philological Society 102(1), 103–129.
Gardner-Chloros. P. & Finnis, K. 2004. How code-switching mediates politeness: genderrelated speech among London Greek-Cypriots. Estudios de Sociolinguística.
Special Issue: Language and Gender: An interdisciplinary perspective. 4(2),
Gardner-Chloros, P., McEntee-Atalianis, L. & Finnis, K. 2005. Language attitudes and
use in a transplanted setting: Greek Cypriots in London. International Journal of
Multilingualism 52–80.
Gardner-Chloros, P., Moyer, M. & Sebba, M. 2007. The LIDES Corpus. In J. C. Beal,
K. P. Corrigan & H. Moisl (eds), Models and Methods in the Handling of
Unconventional Digital Corpora. Vol. 1: Synchronic Corpora; Vol.2: Diachronic
Corpora. Basingstoke: Palgrave, 91–121.
Gardner-Chloros, P., Moyer, M., Sebba, M. & van Hout, R. 1999. Towards standardizing
and sharing bilingual data. International Journal of Bilingualism 3(4), 395–424.
Garrett, M. 1980. Levels of processing in sentence production. In B. Butterworth (ed.),
Language Production, Vol. 1. New York: Academic Press, 170–220.
Garrett, P., Griffiths, Y., James, C. & Scholfield, P. 1994. Use of the mother-tongue in
second language classrooms. Journal of Multilingual and Multicultural
Development 15(5), 371–383.
Gawlitzek-Maiwald, I. 2000. I want a chimney builden; the acquisition of infinitival
constructions in bilingual children. In Döpke, 2000, 123–148.
Gawlitzek-Maiwald, I. & Tracy, R. 1996. Bilingual bootstrapping. Linguistics 34: 901–26.
Geluykens, R. 1992. From Discourse Process to Grammatical Construction: on leftdislocation in English. Amsterdam: John Benjamins.
Genesee, F. 1989. Early bilingual language development: one language or two? Journal
of Child Language 16, 161–79. Reprinted in Li Wei, 2000, 327–347.
2001. Bilingual first language acquisition: exploring the limits of the language faculty.
Annual review of Applied Linguistics 21, 153–168.
Genesee, F., Nicoladis, E. & Paradis, M. 1995. Language differentiation in early
bilingual development. Journal of Child Language 22, 611–631.
George, A. 1990. Whose language is it anyway? Some notes on idiolects. The
Philosophical Quarterly 40(160), 275–295.
Giacalone Ramat, Anna 1995. Code-switching in the context of dialect-standard language relations. In Milroy & Muysken, 1995, 45–67.
Giancarli, P.-D. 1999. Fonctions de l’alternance des langues chez des enfants bilingues
francophones- anglophones. Langage et société 88, 59–88.
Gibbons, J. 1987. Code-mixing and Code-choice: a Hong Kong case study. Clevedon:
Multilingual Matters.
Giesbers, H. 1989. Code-switching tussen dialect en standaardtaal. Amsterdam:
P. Meertensinstituut.
Giles, H. & Coupland, N. 1991. Language: contexts and consequences. Milton Keynes:
Open University Press.
Giles, H. & Smith, P. 1979. Accommodation Theory: optimal levels of convergence. In
H. Giles & R. St. Clair (eds), Language and Social Psychology. Oxford: Blackwell,
Goffman, E. 1979. Footing. Semiotica 25, 1–29.
Goglia, F. 2006. Communicative strategies in the Italian of Igbo-Nigerian Immigrants in
Padova, Italy: a Contact-Linguistic Approach. Unpublished PhD thesis, University
of Manchester.
Goodz, N. 1989. Parental language mixing in bilingual families. Journal of Infant Mental
Health 10(1), 25–44.
1994. Interaction between parents and children in bilingual families. In F. Genesee
(ed.), Educating Second Language Children. New York: Cambridge University
Press, 61–81.
Grammont, M. 1902. Observations sur le langage des enfants. Paris: Mélanges Meillet.
Green, D. W. 1986/2000. Control, activation and resource: a framework and a model for
the control of speech in bilinguals. Brain and Language 27, 210–223. Reprinted in
Li Wei, 2000, 407–420.
1993. Towards a model of L2 comprehension and production. In Schreuder &
Weltens, 1993, 249–279.
2001 (guest ed.). The cognitive neuroscience of bilingualism. Special Issue,
Bilingualism, Language and Cognition 4(2).
Greenberg, J. H. 1963. Some universals of grammar with particular reference to the order
of meaningful elements. In J. H. Greenberg (ed.), Universals of Language.
Cambridge: MIT Press, 58–90.
Grice, P. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press.
Grosjean, F. 1982. Life with two Languages. Cambridge, MA: Harvard University Press.
1988. Exploring the recognition of guest words in bilingual speech. Language and
Cognitive Processes 3(3), 233–274.
1989. Neurolinguists, beware! The bilingual is not two monolinguals in one person.
Brain and Language 36, 3–15.
1997. Processing mixed language: issues, findings and models. In De Groot & Kroll,
1997, 225–254.
1998. Studying bilinguals: methodological and conceptual issues. Bilingualism,
Language and Cognition 1(2), 131–149.
2000. Processing mixed language: issues, findings, and models. In Li Wei, 2000,
2001. The bilingual’s language modes. In Nicol, 2001, 1–23.
Grosjean, F, Li, P., Münte, T. F. & Rodriguez-Fornells, A. 2003. Imaging bilinguals:
when the neurosciences meet the language sciences. Bilingualism, Language and
Cognition 6(2), 159–165.
Grosjean, F. & Miller, J. 1994. Going in and out of languages: an example of bilingual
flexibility. Psychological Science 5, 201–206.
Grosjean, F. & Soares, C. 1986. Processing mixed language: some preliminary findings.
In J. Vaid (ed.), Language Processing in Bilinguals: psycholinguistic and neurolinguistic perspectives. Hillsdale, NJ: Lawrence Erlbaum Associates, 145–179.
Gumperz, J. J. 1964. Hindi–Panjabi code-switching in Delhi. In H. Lunt (ed.),
Proceedings of the Ninth International Congress of Linguists, Boston, MA. The
Hague: Mouton de Gruyter, 1115–1124.
1967. On the linguistic markers of bilingual communication. Journal of Social Issues
23(2), 48–57.
1982/1991. Discourse Strategies, 2nd ed. Cambridge: Cambridge University Press.
Gumperz, J. & Hernandez, E. 1969. Cognitive aspects of bilingual communication.
Working Paper No.28, Language Behavior Research Laboratory, December 1969.
Berkeley: University of California Press.
Gumperz, J. J. & Hymes, D. 1972 (eds). Directions in Sociolinguistics. New York: Holt
Rinehart & Winston.
Gumperz, J. J. & Wilson, R. D. 1971. Convergence and creolization: a case from the
Indo-Aryan-Dravidian border. Hymes, 1971, 151–168.
Gut, U. 2000. Cross-linguistic structures in the acquisition of intonational phonology by
German–English bilingual children. In Döpke, 2000, 201–225.
Gutierrez, M. J. & Silva-Corvalan, C. 1992. Spanish clitics in a contact situation. In
A. Roca & J. Lipski (eds), Spanish in the United States. Linguistic Contact and
Diversity. Berlin: Mouton de Gruyter, 75–89.
Guttfreund, D. 1990. Effects of language use on the emotional experience of
Spanish–English and English–Spanish bilinguals. Journal of Consulting and
Clinical Psychology 58(5), 604–607.
Halmari, H. 1997. Government and Codeswitching: explaining American Finnish.
Amsterdam/Philadelphia: John Benjamins.
Halmari, H. & Smith, W. 1994. Code-switching and register shift: evidence from
Finnish–English child bilingual conversation. Journal of Pragmatics 21, 427–445.
Hamers, J. & Blanc, M. H. A. 2000. Bilinguality and Bilingualism, 2nd ed. Cambridge:
Cambridge University Press.
Harris, A. C. & Campbell, L. 1995. Historical Syntax in Cross-Linguistic Perspective.
Cambridge: Cambridge University Press.
Hasselmo, N. 1972. Code-switching as ordered selection. In. E. Scherabon Firchow,
K. Grimstad, N. Hasselmo & W. A. O’Neil (eds), Studies for Einar Haugen. The
Hague: Mouton de Gruyter, 261–280.
Haugen, E. 1950. The analysis of linguistic borrowing. Language 26(2), 210–231.
1956. Bilingualism in the Americas: a Bibliography and Research Guide. Publications
of the American Dialect Society 26, Alabama: University of Alabama Press.
Haust, D. 1995. Codeswitching in Gambia: eine soziolinguistische Untersuchung von
Mandinka, Wolof und Englisch in Kontakt. Cologne: Köppe Verlag.
Haust, D. & Dittmar, N. 1998. Taxonomic or functional models in the description of
codeswitching? Evidence from Mandinka and Wolof in African contact situations.
In Jacobson, 1998a, 79–90.
Heath, J. 1989. From Code-switching to Borrowing: foreign and diglossic mixing in
Moroccan Arabic. London/New York: Kegan Paul International.
Heine, B. & Kuteva, T. 2005. Language Contact and Grammatical Change. Cambridge:
Cambridge University Press.
Heller, M. 1988a (ed.). Codeswitching: anthropological and sociolinguistic perspectives. Berlin: Mouton de Gruyter.
1988b. Strategic ambiguity: code-switching in the management of conflict. In Heller,
1988a, 77–96.
1992. The politics of code-switching and language choice. Journal of Multilingual
and Multicultural Development 13(1–2), 123–42.
Heller, M. & Martin-Jones, M. 2001 (eds). Voices of Authority. Education and Linguistic
Difference. Westport, CT: Ablex Publishing.
Herdina, P. & Jessner, J. 2002. A Dynamic Model of Multilingualism: perspectives of
change in psycholinguistics. Clevedon: Multilingual Matters.
Hewitt, R. 1986. White Talk Black Talk. Cambridge: Cambridge University Press.
Hill, J. & Hill, K. 1986. Speaking Mexicano: dynamics of syncretic language in Central
Mexico. Tucson: University of Arizona Press.
Hinrichs, L. 2006 (ed.). Codeswitching on the Web: English and Jamaican creole in
e-mail communication (Pragmatics and Beyond). Amsterdam: John Benjamins.
Hoffmann, C. 1991. An Introduction to Bilingualism. London: Longman.
2001. Towards a description of trilingual competence. International Journal of
Bilingualism 5, 1–17.
Højrup, T. 1983. The concept of life-mode: a form-specifying mode of analysis applied
to contemporary Western Europe. Ethnologia Scandinavica, 15–50.
Holmes, J. 1995. Women, Men and Politeness. Harlow: Longman.
Hudson, L. 1966. Contrary Imaginations. Harmondsworth: Penguin.
Hulk, A. and Müller, N. 2000. Bilingual first language acquisition at the interface
between syntax and semantics. Bilingualism: Language and Cognition 3, 227–244.
Hurtado, N. M. 2003. Code-switching as a communicative and conceptual tool in the
classroom discourse of bilingual students. Unpublished University of California at
Santa Barbara dissertation. Dissertation Abstracts International, A: the Humanities
and Social Sciences, 63, 9, March, 3091-A.
Huwaë, R. 1992. Tweetaligheid in Wierden: het taalgebruik van jongeren uit een
Molukse gemeenschap. Unpublished MA thesis in linguistics, Universiteit van
Hyltenstam, K. and Obler, L. K. 1989 (eds). Bilingualism across the Lifespan.
Cambridge: Cambridge University Press.
Hyltenstam, K. & Stroud, C. 1989. Bilingualism in Alzheimer’s dementia: two case
studies. In Hyltenstam & Obler, 1989, 202–227.
Hymes, D. 1971 (ed.). Pidginization and Creolization of Languages. Cambridge:
Cambridge University Press.
Jackendoff, R. 1997. The Architecture of the Language Faculty. Cambridge, MA: MIT
Jacobson, R. 1998a (ed.). Codeswitching Worldwide. Berlin: Mouton de Gruyter.
1998b. Conveying a broader message through bilingual discourse. An attempt at
contrastive codeswitching research. In Jacobson 1998a, 51–76.
2001 (ed.). Codeswitching Worldwide II. Berlin: Mouton de Gruyter.
Jake, J. L. 1998 Constructing Interlanguage: building a composite Matrix Language.
Linguistics 36, 333–382.
Jake, J. L., Myers-Scotton, C. & Gross, S. 2002. Making a Minimalist approach to
codeswitching work: adding the Matrix Language. Bilingualism, Language and
Cognition 5(1), 69–91.
Jakobson, R., Fant, M. & Halle, M. 1952. Preliminaries to Speech Analysis: the distinctive features and their correlates. Cambridge, MA: MIT Press. Reprinted in
Selected Writings of Roman Jakobson. Major Works 1976–1980. Vol.8. Berlin/
NY/Amsterdam: Mouton de Gruyter, 583–646.
Janički, K. 1990. Towards Non-essentialist Sociolinguistics. Berlin: Mouton de Gruyter.
Javier, R. A. & Marcos, L. R. 1989. The role of stress on language-independence and
code-switching phenomena. Journal of Psycholinguistic Research 18(5), 449–472.
Jespersen, O. 1928. Growth and Structure of the English Language, 4th ed. Leipzig:
Jisa, H. 2000. Language mixing in the weak language: evidence from two children.
Journal of Pragmatics 32(9), 1363–1386.
Johanson, L. 2002. Structural Factors in Turkic Language Contacts. Richmond: Curzon
Johnson, C. & Lancaster, P. 1998. The development of more than one phonology: a
case study of a Norwegian–English bilingual child. International Journal of
Bilingualism 2(3), 265–300.
Johnson, J. S. & Newport, E. L. 1989. Critical period effects in second language learning:
the influence of maturational state on the acquisition of English as a second
language. Cognitive Psychology 21, 60–99.
Jones, M. 1996. Foundations of French Syntax. Cambridge: Cambridge University Press.
1998. Language Obsolescence and Revitalization: linguistic change in two sociolinguistically contrasting Welsh communities. Oxford: Oxford University Press.
2005. Some structural and social correlates of single word intrasentential codeswitching in Jersey Norman French. Journal of French Language Studies 15, 1–23.
Jǿrgensen, J. N. 2005. Plurilingual conversations among bilingual adolescents. Journal
of Pragmatics 37, 391–402.
Joshi, A. K. 1985. Processing of sentences with intrasentential code-switching. In
D. R. Dowty, L. Karttunen & A. M. Zwicky (eds), Natural Language Processing:
psychological, computational and theoretical perspectives. Cambridge: Cambridge
University Press, 190–205.
Juan-Garau, M. & Pérez-Vidal, C. 2001. Mixing and pragmatic parental strategies in
early bilingual acquisition. Journal of Child Language 28, 56–86.
Kamwangamalu, N. & Li, C. 1991. Chinese–English code-mixing: a case of matrix
language assignment. World Englishes 10, 247–261.
Karousou-Fokas, R. & Garman, M. 2001. A psycholinguistic interpretation of codeswitching: evidence from fluent Greek-English bilingual adults. International
Journal of Bilingualism 5(1), 39–71.
Kecskes, I. & Moyer, M. 2002. An ethno-conceptual approach to codeswitching. Paper
presented at Sociolinguistics Symposium 14, University of Ghent, Belgium, April
Kerswill, P. 1994. Dialects Converging: rural speech in urban Norway. Oxford: Oxford
University Press.
Kessler, C. 1971. The Acquisition of Syntax in Bilingual Children. Washington, DC:
Georgetown University Press.
King, R. 2000. The Lexical Basis of Grammatical Borrowing: a Prince Edward Island
French case-study. Amsterdam/Philadelphia: Benjamins.
Klavans, J. L. 1985. The syntax of code-switching: Spanish and English. Proceedings of
the Linguistic Symposium on Romance Languages. Amsterdam: Benjamins.
Kolers, P. A. 1963. Interlingual word associations. Journal of Verbal Learning & Verbal
Behavior 2, 291–300.
Köppe, R. & Meisel, J. M. 1995. Code-switching in bilingual first language acquisition.
In Milroy & Muysken, 1995, 276–302.
Kroll, J. F. 1993. Accessing conceptual representations for words in a second language.
In Schreuder & Weltens, 1993, 53–83.
Labov, W. 1971. The notion of ‘system’ in creole languages. In Hymes, 1971, 447–472.
1972. Sociolinguistic Patterns. Philadelphia, PA: University of Pennsylvania
Lanvers, U. 2001. Language alternation in infant bilinguals: a developmental approach
to codeswitching. International Journal of Bilingualism 5(4), 437–464.
Lanza, E. 1992. Can bilingual two-year-olds code-switch? Journal of Child Language
19, 633–658.
1997. Language Mixing in Infant Bilingualism: a sociolinguistic perspective. Oxford:
Oxford University Press.
2001. Bilingual first language acquisition: a discourse perspective on language contact
in parent–child interaction. In Cenoz & Genesee, 2001, 201–229.
Lavandera, B. 1978. The variable component in bilingual performance. In J. E. Alatis
(ed.), International Dimensions of Bilingual Education. Washington DC:
Georgetown University Press, 391–409.
Lawson-Sako, S. & Sachdev, I. 1996. Ethnolinguistic communication in Tunisian streets.
In Y. Suleiman (ed.), Language and Ethnic Identity in the Middle East and North
Africa. Richmond: Curzon Press, 61–79.
2000. Codeswitching in Tunisia: attitudinal and behavioural dimensions. Journal of
Pragmatics 32, 1343–1361.
Lebrun, Y. 1991. Polyglotte Reaktionen. Neurolinguistik, 5, 1–9.
Leech, G., Myers, A. & Thomas, J. A. 1995 (eds). Spoken English on Computer. London:
Leopold, W. 1939–1949. Speech Development of a Bilingual Child: a linguist’s record.
4 vols. Evanston, IL: Northwestern University Press.
Le Page, R. B. 1989. What is a Language? York Papers in Linguistics 13, 9–24.
1997. The evolution of a sociolinguistic theory of language. In Coulmas, 1997, 15–32.
Le Page, R, and Tabouret-Keller, A. 1985. Acts of Identity. Cambridge: Cambridge
University Press.
Levelt, W. J. M. 1989. Speaking: from intention to articulation. Cambridge, MA: MIT
Lin, A. M. Y. 2001. Symbolic domination and bilingual classroom practices in Hong
Kong schools. In Heller & Martin-Jones, 2001, 139–168.
Lindholm, K. A. & Padilla, A. M. 1978. Language mixing in bilingual children Journal
of Child Language 5, 327–335.
LIPPS Group 2000. The LIDES coding manual: a document for preparing and analysing
language interaction data. Special Issue, International Journal of Bilingualism.
Lipski, J. M. 1978. Code-switching and the problem of bilingual competence. In Paradis,
1978, 250–264.
Lisker, L. & Abramson, A. 1964. A cross-language study of voicing in initial stops. Word
20, 384–422.
Li Wei 1994. Three Generations, Two Languages, One Family: language choice and
language shift in a Chinese community in Britain. Clevedon: Multilingual Matters.
1998a. Banana split? Variations in language choice and code-switching patterns of two
groups of British-born Chinese in Tyneside. In Jacobson, 1998a, 153–176.
1998b. The ‘why’ and ‘how’ questions in the analysis of conversational codeswitching. In Auer, 1998b, 156–180.
2000 (ed.). The Bilingualism Reader. London: Routledge.
2002. ‘What do you want me to say?’ On the Conversation Analysis approach to
bilingual interaction. Language in Society 31, 159–180.
2005. “How can you tell?” Towards a common sense explanation of conversational
code-switching. Journal of Pragmatics 37(3), 375–389.
Li Wei, Milroy, L. & Pong Sin Ching 2000. A two-step sociolinguistic analysis of codeswitching and language choice: the example of a bilingual Chinese community in
Britain. In Li Wei, 2000, 188–209.
Lüdi, G. 1993. Second language acquisition by migrants in Europe. In R. Posner &
J. N. Green (eds), Trends in Romance Linguistics and Philology. Berlin: Mouton de
Gruyter, 495–534.
2003. Code-switching and unbalanced bilingualism. In Dewaele, Housen & Li Wei,
2003, 174–188.
Macnamara, J. 1971. The bilingual’s linguistic performance: a psychological overview.
Journal of Social Issues 23(2), 67–71.
MacSwan, J. 1999. A Minimalist Approach to Intrasentential Code Switching. New
York: Garland.
2000. The architecture of the bilingual language faculty: evidence from codeswitching. Bilingualism, Language and Cognition 3(1), 37–54.
2004. Code switching and linguistic theory. In Bhatia & Ritchie, 2004, 283–312.
2005. Code-switching and generative grammar: a critique of the MLF model and
some remarks on ‘modified minimalism’. Bilingualism, Language and Cognition
8(1), 1–22.
MacWhinney, B. 1995 [1991]. The CHILDES Project: Tools for Analyzing Talk. 2nd ed.
Hillsdale, NJ: Erlbaum.
2007. The TalkBank Project. In Joan C. Beal, Karen P. Corrigan & Hermann Moisl
(eds), Creating and Digitizing Language Corpora: synchronic databases
(Volume 1), 163–180. London: Palgrave.
MacWhinney, B. & Snow, C. 1990. The Child Language Data Exchange System: an
update. Journal of Child Language 17, 457–472.
Maehlum, B. 1990. Codeswitching in Hemnesberget–Myth or Reality? In E. Hakon
Jahr & O. Lorentz (eds), Tromsø Linguistics in the Eighties. Oslo: Novus Press,
338–355. Reprinted in Journal of Pragmatics 1996, 25(6), 749–761.
Mahootian, S. 1993. A null theory of codeswitching. Unpublished doctoral dissertation,
Northwestern University.
2005. Written code: the discourse of social change. Talk presented at ISB 5, Barcelona,
March 2005.
Maneva, B. & Genesee, F. 2002. Bilingual babbling: evidence for language differentiation
in dual language acquisition. In Proceedings of the 26th Boston University
Conference on Language Development. Somerville, MA: Cascadilla Press.
Mangruel, A. 2004. Journal d’un lecteur. Translated from English by Christine Le
Boeuf. France: Actes Sud.
Maniakas, T. M. 1991. KANO + INF.: the case of a Greek auxiliary verb in a language
contact situation. Journal of Applied Linguistics 7, 114–131.
Man-Siu Yau, F. 1997. Code switching and language choice in the Hong Kong
Legislative Council. Journal of Multilingual and Multicultural Development
18(1), 40–53.
Marantz, A. 1995. The Minimalist Program. In Gert Webelhuth (ed.), Government and
Binding Theory and the Minimalist Program: principles and parameters in syntactic theory. Oxford: Blackwell, 349–381.
Martin, E. 2002. Mixing English in French advertising. World Englishes 10(2), 139–151.
Martin, P. 2003. Bilingual encounters in the classroom. In Dewaele, Housen & Li Wei,
2003, 67–87.
Martin-Jones, M. 1995. Code-switching in the classroom: Two decades of research. In
Milroy & Muysken, 1995, 90–112.
2000. Bilingual classroom interaction: a review of recent research. Language
Teaching 33, 1–9.
Martin-Jones, M. & Romaine, S. 1986. Semilingualism: a half-baked theory of communicative competence. Applied Linguistics 7(1), 26–38.
Marty, S. & Grosjean, F. 1998. Aphasie, bilinguisme et modes de communication.
Aphasie und verwandte Gebiete 12(1), 8–28.
Maters, K. 1979. An evaluation of syntactic constraints on code-switching and their
potential application to Dutch/English. Unpublished MPhil thesis, University of
Matras, Y. 2000a. Mixed languages: a functional-communicative approach.
Bilingualism, Language and Cognition 3(2), 79–99.
2000b. Fusion and the cognitive basis for bilingual discourse markers. Bilingualism,
Language and Cognition 4(4), 505–529.
Matzen, R. 1980. Dichte isch bichte: Gedichte in Strassburger Mundart. Kehl: Morstadt
Mayes, P. 1990. Quotation in spoken English, Studies in Language 14, 325–363.
McClure, E. 1981. Formal and functional aspects of the code switched discourse of
bilingual children. In R. P. Duran (ed.), Latina Language and Communicative
Behaviour. Norwood, NJ: Ablex, 69–94.
1998. The relationship between form and function in written national language –
English codeswitching: evidence from Mexico, Spain and Bulgaria. In Jacobson,
1998a, 125–152.
McClure, E & McClure, M. 1988. Macro- and micro-sociolinguistic dimensions of codeswitching in Vingard (Romania). In Heller, 1988a, 25–51.
McConvell, P. 1988. Mix-im-up: Aboriginal code-switching, old and new. In M. Heller
(ed.) Codeswitching: anthropological and sociolinguistic perspectives. Berlin:
Mouton de Gruyter, 97–124.
McCormick, K. 2002. Language in Cape Town’s District Six. Oxford: Oxford University
McLelland, N. 2004. A historical study of codeswitching in writing: German and Latin in
Schottelius’ Ausführliche Arbeit von der Teutschen Haubtsprache (1663).
International Journal of Bilingualism 8(4), 499–525.
Meeuwis, M. & Blommaert, J. 1994. The ‘markedness model’ and the absence of society.
Multilingua 13, 387–423.
1998. A monolectal view of code-switching: layered code-switching among Zairians
in Belgium. In Auer, 1998c, 76–101.
Meisel, J. M. 1989. Early differentiation of languages in bilingual children. In
Hyltenstam & Obler, 1989, 13–41. Reprinted in Li Wei, 2000, 344–369.
1994 (ed.). Bilingual First Language Acquisition: French and German grammatical
development. Amsterdam/Philadelphia: Benjamins.
2000. Early differentiation of languages in bilingual children. In Li Wei, 2000,
Merritt, M., Cleghorn, A., Abagi, J. O. & Bunyi, G. 1992. Socializing multilingualism:
determinants of codeswitching in Kenyan primary classrooms. Journal of
Multilingual and Multicultural Development 13 (1/2), 103–122.
Meuter, R. F. I. 2005. Language selection in bilinguals: mechanisms and processes. In
J. F. Kroll & A. M. B. de Groot (eds), The Handbook of Bilingualism: psycholinguistic approaches. Oxford: Oxford University Press, 349–370.
Michina, S. 1999. The role of parental input and discourse strategies in the early language
mixing of a bilingual child. Multilingua 18(4), 317–342.
Milroy, J. 1998. Towards a speaker-based account of language change. In E. Håkon Jahr
(ed.), Language Change: advances in historical sociolinguistics. Berlin: Mouton de
Gruyter, 21–36.
Milroy, L. & Gordon, M. 2003. Sociolinguistics: method and interpretation. Oxford:
Milroy, L. & Li Wei 1995. A social network approach to code-switching: the example of
a bilingual community in Britain. In Milroy & Muysken, 1995, 136–157.
Milroy, L. & Muysken, P. 1995 (eds). One Speaker, Two Languages: cross-disciplinary
perspectives on code-switching. Cambridge: Cambridge University Press.
Mishina, S. 1999. The role of parental input and discourse strategies in the early language
mixing of a bilingual child. Multilingua 18(4), 317–342.
Mishoe, M. 1998. Styleswitching in Southern English. In Myers-Scotton, 1998,
Mitchell, R. & Miles, F. 2004. Second Language Learning Theories, 2nd ed. London:
Montes-Alcala, C. 1998. Written codeswitching: powerful bilingual images. In
Jacobson, 2001, 193–223.
Moreno, E. M., Federmaier, K. D. & Kutas, M. 2002. Switching languages, switching
palabras (words): an electrophysiological study of code-switching. Brain and
Language 80(2), 188–207.
Morita, E. 2003. Children’s use of address and reference terms. Multilingua 22, 367–395.
Morton, J. 1979. Word recognition. In J. Morton & J. Marshall (eds), Psycholinguistics:
Series 2. Structures and Processes. London: Elek, 107–156.
Mougeon, R. & Beniak, E. 1991. Linguistic Consequences of Language Contact and
Restriction: The case of French in Ontario, Canada. Oxford: Oxford University
Moyer, M. G. 1992. Analysis of codeswitching in Gibraltar. Unpublished PhD dissertation, Universitat Autònoma de Barcelona.
1998. Bilingual conversation strategies in Gibraltar. In Auer, 1998c, 215–236.
2000. Gibraltar oral and written corpus 1987–1990. Unpublished.
Muñoz, M. C. 1999. Usages du français et double appartenance: le cas des Portugais en
France. Education et Sociétés Plurilingues, 7, December, 21–33.
Muysken, P. 1990. Concepts, methodology and data in language research: ten remarks
from the perspective of grammatical theory. Papers for the workshop on concepts,
methodology and data. European Science Foundation Netword on Code-Switching.
1995. Code-switching and grammatical theory. In Milroy & Muysken, 1995, 177–199.
1997. Accounting for variation in code-mixing patterns. Paper presented at the
Bilingualism Symposium, Newcastle, United Kingdom.
2000. Bilingual Speech: a typology of code-mixing. Cambridge: Cambridge
University Press.
Muysken, P., Kook, H. & Vedder, P. 1996. Papiamento/Dutch code-switching in bilingual parent–child reading. Applied Psycholinguistics 17, 485–505.
Myers-Scotton, C. 1983. The negotiation of identities in conversation: a theory of
markedness and code choice. International Journal of the Sociology of Language
44, 115–136.
1986. Diglossia and code-switching. In J. A. Fishman, A. Tabouret-Keller, M. Clyne,
B. Krishnamurti & M. Abdulaziz (eds), The Fergusonian Impact, 2 vols, Berlin:
Mouton de Gruyter, ii, 403–417.
1988. Code-switching as indexical of social negotiations. In Heller, 1988a, 151–186.
1992. Comparing code-switching and borrowing. Journal of Multilingual and
Multicultural Development 13(1/2), 19–39.
1993a. Social Motivations for Code-switching: evidence from Africa. Oxford: Oxford
University Press.
1993b Duelling Languages. Oxford: Oxford University Press.
1997. Codeswitching. In Coulmas, 1997, 217–237.
1998. (ed.) Codes and Consequences: choosing linguistic varieties. Oxford: Oxford
University Press.
1999. Explaining the role of norms and rationality in codeswitching. Journal of
Pragmatics 32, 1259–1271.
2002a. Contact Linguistics: bilingual encounters and grammatical outcomes. Oxford:
Oxford University Press.
2002b. Frequency and intentionality in (un)marked choices in CS: ‘This is a 24 hour
country’. International Journal of Bilingualism 6(2), 205–219.
2003. Code-switching: evidence of both flexibility and rigidity in language. In
Dewaele, Housen & Li Wei, 2003, 189–203.
Myers-Scotton, C. & Bolonyai, A. 2001. Calculating speakers: codeswitching in a
Rational Choice Model. Language in Society 30, 1–28.
Myers-Scotton, C. & Jake, J. L. 1995/2000. Matching lemmas in a bilingual competence and
production model. Linguistics 33, 981–1024. Reprinted in Li Wei, 2000, 281–320.
2000. Testing the 4-M model: an introduction. International Journal of Bilingualism
4(1), 1–8.
2001. Explaining aspects of code-switching and their implications. In Nicol, 2001,
2009. No chaos allowed. In B. Bullock & J. Toribio. (eds), Handbook of Codeswitching. Cambridge: Cambridge University Press.
Nehrlich, B. & Clarke, D. D. 1994. Language, action and context: linguistic pragmatics
in Europe and America (1800–1950). Journal of Pragmatics 22, 439–463.
Neufeld, G. G. 1976. The bilingual’s lexical store, International Review of Applied
Linguistics 14(1), 15–35.
Nicol, J. L. 2001. One Mind, Two Languages: bilingual language processing. Oxford:
Nicoladis, E. & Secco, G. 2000. The role of a child’s productive vocabulary in the
language choice of a bilingual family. First Language 20, 3–28.
Nilep, C. 2006. ‘Code switching’ in sociocultural linguistics. Colorado Research in
Linguistics 19, 1–68.
Nishimura, M. 1997. Japanese/English code-switching: syntax and pragmatics. Berkeley
Insights in Linguistics and Semiotics 24, New York: Peter Lang.
Nivens, R. J. 2002. Borrowing versus Code-switching in West Tarangan (Indonesia).
Dallas, TX: SIL International.
Nortier, J. 1990. Dutch–Moroccan Arabic Code Switching. Dordrecht: Foris Publications.
Obler, L. K., Zatorre, R. J., Galloway, L. & Vaid, J. 2000. Cerebral lateralization in
bilinguals: methodological issues. In Li Wei, 2000, 381–394.
Ochs, E. 1979. Transcription as theory. In E. Ochs & B. B. Schieffelin (eds),
Developmental Pragmatics. New York: Academic Press, 43–72.
Oesch-Serra, C. 1998. Discourse connectives in bilingual conversation: the case of an
emerging Italian–French mixed code. In Auer, 1998c, 101–124.
Oezguezel, S. 1994. De vitaliteit van het Turks in Nederland. Unpublished PhD dissertation, Tilburg University.
Otheguy, R. 1995. When contact speakers talk, linguistic theory listens. In E. ContiniMrava & B. Sussman Goldberg (eds), Advances in Linguistic Sign Theory. Berlin:
Mouton de Gruyter, 213–242.
Pandharipande, R. V. 1998. Is genetic connection relevant in code-switching? Evidence
from South Asian languages. In Jacobson, 1998a, 201–220.
Pandit, I. 1990. Grammaticality in code-switching. In R. Jacobson (ed.), Codeswitching
as a Worldwide Phenomenon. New York: Peter Lang, 33–69.
Paradis, J. 2000. Beyond ‘One system or two’? Degrees of separation between the
languages of French–English bilingual children. In Döpke, 2000, 175–201.
Paradis, J., Nicoladis, E. & Genesee, F. 2000. Early emergence of structural constraints
on code-mixing: evidence from French–English bilingual children. Bilingualism,
Language and Cognition 3(3), 245–261.
Paradis, M. 1978 (ed.). Aspects of Bilingualism. Columbia, SC: Hornbeam Press.
1981. Neurolinguistic organization of a bilingual’s two languages. In J. E. Copeland &
P. W. Davis (eds), The 7th LACUS Forum. Columbia, SC: Hornbeam Press,
1987. The Assessment of Bilingual Aphasia. Hillsdale, NJ: Lawrence Erlbaum.
1995. Selective deficit in one language is not a demonstration of a different anatomical representation: comments on Gomez-Tortosa et al. Brain & Language 54,
Paradis, M. & Goldblum, M. C. 1989. Selective crossed aphasia in a trilingual aphasic
patient followed by reciprocal antagonism. Brain & Language 36, 62–75.
Pavlenko, A. 2004. ‘Stop doing that, ia komu skazala!’: language choice and emotions
in parent–child communication. Journal of Multilingual & Multicultural
Development 25(2/3), 179–203.
2005. Emotions and Multilingualism. Cambridge: Cambridge University Press.
Peal, E. & Lambert, W. 1962. The relationship of bilingualism to intelligence.
Psychological Monographs 76, 1–23.
Penfield, W. & Roberts, L. 1959. Speech and Brain Mechanisms. Princeton, NJ:
Princeton University Press.
Perecman, E. 1984. Spontaneous translation and language mixing in a polyglot aphasic.
Brain and Language 23, 43–63.
1989. Language processing in the bilingual: evidence from language mixing. In
Hyltenstam & Obler, 1989, 227–245.
Perrot, M.-E. 2001. Ordre des mots et restructurations dans le chiac de Moncton:
l’example du syntagme nominal. In Canut & Caubet, 2001, 49–59.
Pfaff, C. 1979. Constraints on language-mixing: intrasentential code-switching and
borrowing in Spanish–English. Language 55, 291–318.
Pinker, S. 1994. The Language Instinct. Harmondsworth: Penguin.
Poplack, S. 1980. Sometimes I’ll start a sentence in Spanish Y TERMINO EN
ESPAÑOL: toward a typology of code-switching. Linguistics 18, 581–618.
1988. Contrasting patterns of code-switching in two communities, in Heller, 1988a,
2000. Preface to the reprint of ‘Sometimes I’ll start a sentence in Spanish Y
TERMINO EN ESPAÑOL: toward a typology of code-switching’ (see Poplack
1980). In Li Wei, 2000, 221–223.
Poplack, S. & Meechan, M. 1995. Patterns of language mixture: nominal structure in
Wolof–French and Fongbe–French bilingual discourse. In Milroy & Muysken,
1995, 199–232.
Poplack, S. & Sankoff, D. 1984. Borrowing: the synchrony of integration. Linguistics 22,
Poplack, S., Sankoff, D. & Miller, C. 1988. The social correlates and linguistic processes
of lexical borrowing and assimilation. Linguistics 26, 47–104.
Poplack, S., Wheeler, S. & Westwood, A. 1987. Distinguishing language contact
phenomena: evidence from Finnish–English bilingualism. In P. Lilius & M. Saari
(eds), The Nordic Languages and Modern Linguistics, 6, Helsinki: University of
Helsinki Press, 33–56. Reprinted in R. Jacobson (ed.) 1990, Code-switching as a
World-wide Phenomenon, 170–185. New York: Peter Lang.
Popper, K. 1959. The Logic of Scientific Discovery. London: Hutchinson.
Pötzl, O. 1925. Űber die parietal bedingte Aphasie und ihren Einfluss auf das Sprechen
mehrerer Sprachen [On aphasia of parietal origin and its influence on the speaking
of several languages]. Zeitshrift für die gesamte Neurologie und Psychiatrie 96,
100–124. Translated in M. Paradis (ed.), 1983, Readings on Aphasia in Bilinguals
and Polyglots. Montreal; Marcel Didier, 176–198.
Poulisse, N. 1997. Language production in bilinguals. In De Groot & Kroll, 1997, 201–225.
Poulisse, N. & Bongaerts, T. 1994. First language use in second language production.
Applied Linguistics 15, 36–57.
Pujadas, J., Pujol Berché, M. & Turell, M. T. 1988–1992. Catalan–Spanish database.
Pujolar i Cos, J. 2001. Gender, Heteroglossia and Power. A sociolinguistic study of youth
culture. Berlin/New York: Mouton de Gruyter.
Radford, A. 1996. Towards a structure-building model of acquisition. In H. Clahsen
(ed.), Generative Perspectives on Language Acquisition. Amsterdam/Philadelphia:
John Benjamins, 43–91.
Ramat, A. G. 1995. Code-switching in the context of dialect/standard language relations.
Milroy & Muysken, 1995, 45–67.
Rampton, B. 1995. Crossing: language and ethnicity among adolescents. London:
Redlinger, W. E. & Park, T. 1980. Language mixing in young bilinguals. Journal of
Child Language 7, 337–352.
Rindler-Schjerve, R. 1998. Code-switching as an indicator for language shift? Evidence
from Sardinian–Italian bilingualism. In Jacobson, 1998a, 221–248.
Robinson, W. P. & Giles, H. 2001. The New Handbook of Language and Social Psychology.
Chichester, NY/ Weinheim/Brisbane/Singapore/Toronto: John Wiley & Sons.
Roca, A. & Jensen, J. B. 1996. Spanish in Contact: issues in bilingualism. Somerville,
MA: Cascadilla Press.
Romaine, S. 1986. The syntax and semantics of the code-mixed compound verb in
Panjabi–English bilingual biscourse. In D. Tannen & J. Alatis (eds), Language and
Linguistics. The Interdependence of Theory, Data and Application. Washington
DC: Georgetown University Press, 35–49.
1989. Bilingualism. Oxford: Basil Blackwell.
1995. Bilingualism, 2nd ed. Oxford: Blackwell.
1992. Language, Education and Development: urban and rural Tok Pisin in Papua
New Guinea. Oxford: Oxford University Press.
Ronjat, J. 1913. Le développement du langage observé chez un enfant bilingue. Paris:
Rozensky, R. H. & Gomez, M. Y. 1983. Language switching in psychotherapy with
bilinguals: two problems, two models and some case examples. Psychotherapy
Theory, Research and Practice 20, 152–160.
Rubagumya, C. M. 1994 (ed.). Teaching and Researching Language in African
Classrooms: a Tanzanian perspective. Clevedon: Multilingual Matters.
Rubin, E. J. & Toribio, A. J. 1995. Feature-checking and the syntax of language contact.
In J. Amastae, G. Goodall, M. Montalbetti & M. Phinney (eds), Contemporary
Research in Romance Linguistics. Amsterdam: Benjamins, 177–185.
Sachdev, I. & Bourhis, R. Y. 1990. Bi- and multilingual communication. In H. Giles and
P. Robinson (eds), The Handbook of Language and Social Psychology, Chichester,
NY/Weinheim/Brisbane/Singapore/Toronto: John Wiley & Sons, 293–308.
2001. Multilingual communication. In Robinson & Giles, 2001, 407–429.
Sachdev, I. & Giles, H. 2004. Bilingual accommodation. In Bhatia & Ritchie, 2004,
Sankoff, D. 1998. A formal production-based explanation of the facts of code-switching.
Bilingualism, Language and Cognition 1(1), 39–50.
Sankoff, D. & Mainville, S. 1986. Code-switching and context-free grammars.
Theoretical Linguistics 13, 75–90.
Sankoff, D. & Poplack, S. 1981. A formal grammar for code-switching. Papers in
Linguistics 14(1), 3–46.
Sankoff, D., Poplack, S. & Vanniarajan, S. 1990. The case of the nonce loan in Tamil.
Language Variation and Change 2(1), 71–101.
Sankoff, G. 2001. Linguistic outcomes of language contact. In J. Chambers, P. Trudgill &
N. Schilling-Estes (eds), The Handbook of Language Variation and Change.
Oxford: Blackwell, 638–668.
Sarkar, M. & Winer, L. 2006. Multilingual codeswitching in Quebec rap: poetry,
pragmatics and performativity. International Journal of Multilingualism 3(3),
Saunders, G. 1982. Bilingual Children: guidance for the family. Clevedon: Multilingual
Saville-Troike, M. 1982. The Ethnography of Communication: an introduction. Oxford:
Schelletter, C. 2000. Negation as a crosslinguistic structure in a German–English
bilingual child. In Döpke, 2000, 105–121.
Schendl, H. 2002. Linguistic aspects of code-switching in medieval English texts. In
Trotter, 2002, 77–93.
Schmidt, A. 1985. Young People’s Djirbal. Cambridge: Cambridge University Press.
Schreuder, R. & Weltens, B. 1993 (eds). The Bilingual Lexicon. Amsterdam: Benjamins.
Schuffenecker, G. 1981. Mode de vie en Alsace. Institut national de la statistique et des
études économiques. Dernières Nouvelles d’Alsace, 6–15 February.
Schütze, C. T. 1996. The Empirical Basis of Linguistics: grammaticality judgments and
linguistic methodology. Chicago: University of Chicago Press.
Scotton C. M. 1976. Strategies of neutrality: language choice in uncertain situations.
Language 52(4): 919–941.
Seaman, P. D. 1972. Modern Greek and American English. The Hague: Mouton de
Sebba, M. 1993. London Jamaican: Language Systems in Interaction. London/New
York: Longman.
1995a. Code-switching: a problem for transcription and text encoding. In Leech,
Myers & Thomas, 1995, 144–148.
1995b. Towards a bilingual corpus of code-switching data. Network on codeswitching
and language contact: Proceedings of the ESF Summer School on codeswitching
and language contact, Ljouwert/Leeuwaarden, the Netherlands, 14–17 September,
1994, Fryske Akademy, 229–241.
1998. A congruence approach to the syntax of code-switching. International Journal
of Bilingualism 2, 1–20.
2005. Towards a typology and analytical framework for mixed language texts. Talk
delivered at ISB5, Barcelona, March.
Sebba, M. & Wootton, T. 1998. We, they and identity: sequential v. identity-related
explanation in CS. In Auer, 1998c, 262–290.
Sella-Mazi, E. 1997. Language contact today: the case of the Muslim minority in northeastern Greece. International Journal of the Sociology of Language 126, 83–103.
1999. La minorité musulmane turcophone de Grèce: approche sociolinguistique
d’une communauté bilingue. Athens: Trohalia.
Selting, M., Auer, P. Barden, B., Bergmann, J., Cer-Kuhlen, E., Günthner, S., Meier, C.,
Quasthoff, U., Schlobinski, P. & Uhmann. S. 1998. Gesprächsanalystisches
Transkriptionssystem (GAT). Linguistische Berichte 173, 91–122.
Shepard, C. A., Giles, H. & Le Poire, B. A. 2001. Communication Accommodation
Theory. In Robinson & Giles, 2001, 33–56.
Shin, S. J. & Milroy, L. 2000. Conversational code-switching among Korean–English
bilingual children. International Journal of Bilingualism 4(3), 351–383.
Siegel, J. 1994. Review of S. Romaine ‘Language, education and development’.
Language in Society 23, 144–149.
Sifianou, M. 1992. Politeness Phenomena in England and Greece. A cross-cultural
perspective, Oxford: Oxford University Press.
Singh, R. 1983. We, they and us: a note on code-switching and stratification in North
India. Language in Society 12(1), 71–73.
Sinka, I. 2000. The search for cross-linguistic influences in the language of young
Latvian–English bilinguals. In Döpke, 2000, 149–175.
Smith, N. 1989. The Twitter Machine: reflections on language. Oxford: Blackwell.
Smith, Z. 2000. White Teeth. Penguin Books.
Sokolov, J. & Snow, C. 1994 (eds). Handbook of Research in Language Development
Using CHILDES. Hillsdale NJ: Erlbaum.
Spolsky, B. & Cooper, R. L. 1991. The Languages of Jerusalem. Oxford: Oxford
University Press.
Sridhar, S. & Sridhar, K. 1980. The syntax and psycholinguistics of bilingual codeswitching. Canadian Journal of Psychology 34, 407–416.
Stolt, B. 1964 Die Sprachmischung in Luthers Tischreden: Studien zum Problem der
Zweisprachigkeit (Acta Universitatis Stockholmiensis, Stockholmer germanistische Forshungen 4). Stockholm: Almquist & Wiksell.
Stroud, C. 1992. The problem of definition and meaning in code-switching. Text 12(1),
1998. Perspectives on cultural variability of discourse and some implications for codeswitching. In Auer, 1998c, 321–348.
Swain, S. 2002. Bilingualism in Cicero? The evidence of code-switching. In Adams &
Swain, 2002, 128–168.
Swigart, L. 1991. Women and language choice in Dakar: a case of unconscious innovation. Women and Language 15(1), 11–20.
Tabouret-Keller, A. 1969. Le bilinguisme de l’enfant avant 6 ans: étude en milieu
alsacien. Thèse lettres, Université de Strasbourg.
1995. Conclusion. Code-switching as a theoretical challenge. In Milroy & Muysken,
1995, 344–355.
1997. Language and identity. In Coulmas, 1997, 315–326.
Tabouret-Keller, A. & Luckel, F. 1981a. La dynamique sociale du changement linguistique: quelques aspects de la situation rurale en Alsace. International Journal of the
Sociology of Language 29, 51–66.
1981b. Maintien de l’alsacien et adoption du français; éléments de la situation
linguistique en milieu rural en Alsace. Langages 61, 39–62.
Taeschner, T. 1983. The Sun is Feminine: a study on language acquisition in bilingual
children. Berlin: Springer.
Takashi, K. 1990. A sociolinguistic analysis of English borrowings in Japanese advertising texts. World Englishes 9(3), 327–341.
Tamis, A. 1986. The state of the Modern Greek language as spoken in Victoria.
Unpublished PhD thesis, University of Melbourne.
Tannen D. 1989. Talking Voices: repetition, dialogue, and imagery in conversational
discourse. Cambridge/New York: Cambridge University Press.
Thomason, S. G. 1997. Introduction. In S. G. Thomason (ed.), Contact Languages: a
wider perspective. Amsterdam/Philadelphia: Benjamins, 1–9.
2001. Language Contact: an introduction. Edinburgh: Edinburgh University Press.
Thomason, S. G. & Kaufman, T. 1988. Language Contact, Creolization and Genetic
Linguistics. Berkeley: University of California Press.
Timm, L. 1975. Spanish–English code-switching: el porqué y how-not-to. Romance
Philology 28(4), 473–482.
1978. Code-switching in war and peace. In Paradis, 1978, 302–315.
Toribio, A. J. 2004. Convergence as an optimization strategy in bilingual speech:
evidence from code-switching. Bilingualism: Language and Cognition 7(2),
Toribio, A. J. & Rubin, E. J. 1996. Code-switching in generative grammar. In Roca &
Jensen, 1996, 203–226.
Torras, M. C. 1998. Code negotiation and code alternation in service encounters in
Catalonia. Unpublished MA dissertation, Lancaster University.
Tosi, A. 1984. Immigration and Bilingual Education: a case study of movement of population, language change and education within the EEC. Oxford: Pergamon Press.
Tracy, R. 1995. Child language in contact: bilingual language acquisition (English/
German) in early childhood. Habilitationsschrift, University of Tübingen.
2000. Language mixing as a challenge for Linguistics. In Döpke, 2000, 11–36.
Treffers-Daller, J. 1990. Towards a uniform approach to code-switching and borrowing.
Papers for the Workshop on Constraints, Conditions and Models, European Science
Foundation Network on Code-switching in Language Contact. Strasbourg: ESF,
1992. Switching between French and Dutch: (dis)similarities in the switching patterns
of men and women. Paper presented at the International Conference on CodeSwitching, Nijmegen, Holland, June 1992.
1994. Mixing Two Languages, French–Dutch contact in a comparative perspective.
Berlin: Mouton de Gruyter.
1998. Variability in code-switching styles: Turkish–German code-switching patterns.
In Jacobson, 1998a, 177–198.
1999. Borrowing and shift-induced interference: contrasting patterns in French–
Germanic contact in Brussels and Strasbourg. Bilingualism, Language and
Cognition 2, 77–80.
Trotter, D. A. 2002 (ed.). Multilingualism in Later Medieval Britain. Cambridge:
D.S. Brewer.
Trudgill, P. 1972. Sex, covert prestige and linguistic change in the urban British English
of Norwich. Language in Society 1(2), 179–195.
1974. The Social Differentiation of English in Norwich. Cambridge: Cambridge
University Press.
1977. Creolization in reverse: reduction and simplification in the Albanian Dialects of
Greece. Transactions of the Philological Society 7, 32–50.
1986. Dialects in Contact. Oxford: Blackwell.
1997. Acts of conflicting identity: the sociolinguistics of British pop songs pronunciation. In Coupland & Jaworski, 1997, 251–267.
Tsitsipis, L. D. 1998. A Linguistic Anthropology of Praxis and Language Shift:
Arvanitika (Albanian) and Greek in contact. Oxford: Oxford University Press.
Turell, M. T. & Forcadell, M. 1992. Catalan/Spanish/English Corpus. Unpublished.
1996. Transcription and coding proposal for LIDES based on La Canonja
Corpus. Paper presented at LIPPS meeting III, Nijmegen, Max Planck Institute,
April 1996.
1997. Transcription, tagging and coding of bilingual corpora via LIDES. Papers de
l’IULA (Working Paper). Barcelona: Institut Universitari de Lingüística Aplicada,
Universitat Pompeu Fabra.
Vaid, J. & Hull, R. 2002. Re-envisioning the bilingual brain using functional neuroimaging: methodological and interpretative issues. In F. Fabbro (ed.), Advances in
the Neurolinguistics of Bilingualism: essays in honor of Michel Paradis. University
of Udine: Forum, 315–357.
Valdès-Fallis, G. 1977. Code-switching among bilingual Mexican–American women:
towards an understanding of sex-related language alternation. International Journal
of the Sociology of Language 7, 65–72.
Van der Linden, E. 2000. Non-selective access in child bilingualism activation: the
lexicon. In Döpke, 2000, 37–57.
Vihman, M. M. 1985. Language differentiation by the bilingual infant. Journal of Child
Language 12, 297–324.
1998. A developmental perspective on code-switching: conversations between a pair
of bilingual siblings. International Journal of Bilingualism 2(1), 45–84.
Volterra, V. & Taeschner, T. 1978. The acquisition and development of language by
bilingual children. Journal of Child Language 5, 311–326.
Von Raffler Engel, W. 1976. Linguistic and kinesic correlates in code-switching. In
W. C. McCormack & S. A. Wurm (eds), Language and Man: anthropological
issues. The Hague/Paris: Mouton de Gruyter, 229–239.
Vygotsky, L. 1934/1962/1986. Thought and Language (first translation into English
1962; new edition 1986). Cambridge, MA: MIT Press.
Wasow, T. 1985. Topic…comment: the wizards of ling. Natural Language and
Linguistic Theory 3(4), 485–491.
Weinreich, U. 1953. Languages in Contact: findings and problems. New York:
Linguistic Circle of New York. Reprinted as 1964, Languages in Contact. The
Hague, Mouton de Gruyter.
Wei Zhang 2005. Code-choice in bidialectal interaction: the choice between Putonghua
and Cantonese in a radio phone-in program in Shenzhen. Journal of Pragmatics
37(iii), 355–375.
Wernicke, C. 1874/1977. Der aphasische symptomencomplex: Eine psychologische
studie auf anatomischer basis. In G. H. Eggert (ed.), Wernicke’s Works on
Aphasia: a sourcebook and review. The Hague: Mouton de Gruyter, 91–145.
Williams, G. 1992. Sociolinguistics: a sociological critique. London: Routledge.
Winford, D. 2003. An Introduction to Contact Linguistics. Oxford: Blackwell.
Winter, J. & Pauwels, A. 2000. Gender and language contact research in the
Australian context. Journal of Multilingual and Multicultural Development
21(6), 508–522.
Woolard, K. A. 1988. Code-switching and comedy in Catalonia. In Heller, 1988a,
1997. Between friends: gender, peer group structure and bilingualism in urban
Catalonia. Language in Society 26, 533–560.
1999. Simultaneity and bivalency as strategies in bilingualism. Journal of Linguistic
Anthropology 8(1), 3–29.
Woolard, K. & Genovese, N. 2007. Strategic bivalency in Latin and Spanish in Early
Modern Spain. Language in Society 36(4), 487–509.
Woolford, E. 1983. Bilingual code-switching and syntactic theory. Linguistic Inquiry 14,
Yip, V. & Matthews, S. 2000. Syntactic transfer in a Cantonese–English bilingual child.
Bilingualism, Language and Cognition 3(3), 193–208.
Zentella, A. C. 1997. Growing up Bilingual: Puerto Rican children in New York. Oxford/
Malden, MA: Blackwell.
Zimman, L. 1993. Spanish–English code-switching. Unpublished MA dissertation,
Birkbeck, University of London.
Note: References to boxes, figures and tables take the form 6b, 106f, 62t.
Aaho, T. 96
Abramson, A. 139n8
accent 178
accommodation 78–81, 118, 172
activation 100, 121, 129, 130, 132, 139–40,
141, 174
Adams, J. N. et al. 88–90b
addressee specification 79–80
Adone, D. 162
advertising 5, 6b, 21, 67
Afrikaans–English 160
Agnihotri, R. K. 169
Aikhenvald, A. Y. 12, 29, 82
Aitchison, J. 30, 107
Aitsiselmi, F. 79
Albatroz (literary review) 27
Albert, M. L . 121, 124, 127
Alsatian see French–Alsatian
alternating antagonism 124–5
alternation 11, 12, 104–5, 106f, 173, 201
Alvarez-Cáccamo, C. 92, 168
Amazonian communities 29
Ancient Rome 88–90b
Anderson, B. 178
Angermeyer, P. S. 21
Ann, J. 126
Anselmi, G. 16
aphasia see bilingual aphasia
Appel, R. 50
Arabic see Beur FM; Moroccan Arabic–Dutch;
Moroccan Arabic–English; Moroccan
Arabic–French; Tunisian Arabic–French
Arvanitika 24
asides 74–5
Aslanov, C. 20
Atkinson, J. M. 184
attitudes to code-switching 14–16, 81–2,
171–2, 178–80
audience design 64, 78–81, 118
Auer, P. 26, 58, 71–2, 92, 103, 132, 168
Australia 167
Dutch–English 38, 102, 105, 108,
134–5, 143
German–English 108, 110, 143
automatic translation 125
awareness of own code-switching 15, 111
Azuma, S. 111, 119
Backus, A. 23, 37, 105, 108, 111–12, 170, 176
Bader, Y. 154
Baetens Beardsmore, H. 16, 17
Bahasa Indonesia–Dutch 110
Bakhtin, M. 77, 176
Bakker, P. 35
Bantu 37
Barrett, R. 63–4
Barron-Hauwert, S. 144
base language see Matrix Language
Baugh, A. C. 35
BBC Grenglish 49–54
Belazi, H. M. et al. 99, 104
Belgium see French–Dutch (Brussels); Zairian
community; Belgium
Bell, A. 64, 74, 118
Beniak, E. 30
Benoit, E. et al. 176
Benson, E. J. 10
Bentahila, A. 24, 29, 81, 82, 99, 100,
101, 109
Besnier, N. 143
Beur FM 79
Bhatia, T. K. 96
Bialystock, E. 142
Biber, D. et al. 4n4, 107
Bickerton, D. 162
bilingual aphasia 122–6, 173
brain scanning techniques 122–3
emotional factors 123–4
methodological caution 122
network failures 124–6
bilingual compound verbs 107
diglossia 55–6
education 159–60
elite bilinguals 143, 155, 159, 173–4
individual/group/community 152
and L2 learners 17
studies 18
terminology 16
see also bilingual aphasia; children;
models of bilingual production;
Bislama 27
Bister-Broosen, H. 44
Blanc, M. H. A. 35, 142, 144, 146, 147,
150, 158
Blom, J. P. 59, 109
Blommaert, J. 13, 57–8, 70, 141, 168
Boersma, P. 185
Boeschoten, H. 46, 107, 113, 141, 169
Bokamba, E. 13
Bolonyai, A. 152, 166
bonding 85–6
Bongaerts, T. 160–1
borrowing 12, 30–3, 36, 178, 201
children 146
grammatical category 31
morphophonemic integration 31–2
native synonym displacement 32–3
Bosch, L. 150
Botero, C. et al. 139, 178
Bourdieu, P. 42, 82
Bourhis, R. Y. 71
brain localization/lateralization 126–8
brain scanning techniques 122–3
Brazil, D. 112
Breitborde, L. B. 56
“bridge” phenomena 108, 178
Broca, P. 124, 126
Broersma, M. 136
Brown, P. 81, 83, 84
Brulard, I. 153
Buenos Aires 24
Bulgaria 60, 61
Bullock, B. 41
Burt, S. M. 71
Bynon, T. 30
Cameron, D. 55, 87
Camilleri, A. 160
Canada 35
French–English 55, 61, 62t
Cantonese–English 81, 100, 150, 189–90
Cantonese–Putonghua 64
Carr, P. 153
Cashman, H. 87
Catalan–Castilian 55, 80–1, 189
Catalan–English 150
Chambers, J. K. 36, 83
Chan, B. 99–100
Chana, U. 81
Charles, R. 78
CHAT see LIDES Project
Cheng, K. K. Y. 157
Cheshire, J. 31, 60, 82–3, 84, 176
Chiac 35
children 142–64, 166–7, 175
characteristics of bilingual children 143–4
code-mixing 146, 148–50
code-switchers/code-breakers 162–4
code-switching in the classroom 159–60, 179
cognitive effects of code-switching 157–8
constraints 154
critical period hypothesis 142
development v. non-development issues 145
group studies 158–9
one system or two? 147–54
OPOL (One Parent One Language) 144, 150,
155, 162
pragmatic uses of code-switching 157
productive abilities 150–3
receptive abilities 150
role of input 144–5, 154–7
second-language learners 160–1
“single system” hypothesis 151
terminology 146–7
Chinese–English 108
Britain 70, 72–3, 109, 169, 189–90
Hong Kong 29, 81, 159
Singapore 119, 159
see also Cantonese–English;
Chinese–Malay 157
Chomsky, N. 93, 98, 154
CLAN programs see LIDES Project
Clarke, D. D. 106
classroom code-switching 159–60
clausal chunks 111–12
Cleghorn, A. 160
clitic constraint 95
Clyne, M. 12, 16–17, 31, 38, 46, 93, 102,
105, 108, 110, 112, 129, 135, 140, 143,
167–8, 178
code-mixing (CM) 12–13, 104
children 146, 148–50
code-switching (CS) 4n3
characteristics 165–7
common-sense approach 7–9, 19, 165
as compromise strategy 15
definitions 1–4, 10–11, 26, 165–6, 170–1
future research 177–8
fuzziness 167–71
history 4, 11, 13, 20, 24, 56, 178
speakers’ insights 14–16
studies 9–10, 13–14
study “for its own sake” 5, 7
terminology 7, 10–13, 202
as window on speech and language 4–5
cognitive effects 157–8, 176–7
cognitive grammar 94
Colina, S. 98
Comeau, L. et al. 146, 155
comparisons between 60–3
variation between 60
communities of practice 64
competence 2, 4, 43, 165
compromise 15, 78, 108
congruent lexicalization 12, 105, 106f
constraints 95, 112–13, 167, 172, 182
children 154
clitic constraint 95
equivalence constraint 96–7
free morpheme constraint 95–6, 97, 146
functional head constraint 99
interaction 97
constructed dialogue 77, 176–7
convergence 13, 21, 23
conversational code-switching 10, 58–9,
65–90, 172
accommodation 78–81, 118, 172
attitudes 14–16, 81–2
audience design 78–81
children 157
as discourse structuring device 68–73
external symbolism 66–7
gender 82–7
intentionality 77
local effects 72, 166
vs. monolingual conversational moves 74–7
conversational maxims 69, 71
Cooper, R. L. 20
Coulmas, F. 177
Coupland, N. 64, 70–1
Cowie, F. 94
Creole–London English 3, 27, 28,
33–4, 57
Creole–Punjabi 58
creolization 27, 28, 33–4, 162
critical period hypothesis 142
Cromdal, J. 157
crossing 58, 179
Crowley, T. 27, 33
CS see code-switching
Dabène, L. 48
Davies, E. E. 24, 29, 99, 100, 101, 109
Davis, J. 126
De Bot, K. 120, 123, 125, 129, 132–5, 136, 174
De Houwer, A. 142, 151, 153, 154–5, 162, 175
De Mejía, A.-M. 143, 159
De Villiers, J. G. 155
De Villiers, P. A. 155
decreolization 27, 33–4
Dell, G. 129, 161
Denison, N. 25–6, 55
Deuchar, M. 104, 153, 155, 175, 177
Dewaele, J.-M. 15, 160, 176
Di Sciullo, A. M. et al. 97
diachronic continuum 12, 26
dialects 36, 55, 59, 109, 167, 178
Diaz, R. M. 157
diffusion 19
diglossia 55–6, 171
Dijkstra, T. 130, 139, 140, 174–5
directness 86–7
discourse structuring 68–73
CS as “verbal action” 70–3, 172
CS vs. monolingual conversational moves
Markedness Theory 56, 69–70, 171
Rational Choice Model 69–70
disjointed code-switching 107
domains of usage 55–6, 55n3
Döpke, S. 146, 152, 158
Dorian, N. 26
double morphology 108–9
Duranti, A. 77
Dussias, P. E. 137
Dutch–Bahasa Indonesia 110
Australia 38, 102, 105, 108, 134–5, 143
L2 learners 161
United States 151–2
Dutch–English–French 139–40
Dutch–French (Brussels) 32, 37–8, 107, 110
Dutch–German 148, 156
Dutch–Malay 110
Dutch–Moroccan Arabic 93, 98, 140
Dutch–Turkish 105, 108
Dyirbal–English 24, 109
E-language 93
Eastman, C. M. 10
Eckert, P. 64
Eco, U. 20
Edwards, M. 91, 107
Eilers, R. E. et al. 150
Eliasson, S. 31–2
elite bilinguals 143, 155, 159, 173–4
elite languages 23
emotional factors 123–4
encoding 15–16, 121–2, 124–5
English see Afrikaans–English;
Catalan–English; Chinese–English;
Creole–London English; Dutch–English;
Dutch–English–French; Dyirbal–English;
Estonian–English; Finnish–English;
French–English; German–English; Greek
Cypriots in London; Greek–English;
Gurindji–English; Hebrew–English;
Hindi–English; Hungarian–English;
Italian–English; Japanese–English;
Korean–English; Latvian–English;
Malay–English; Mandinka–
Wolof–English; Marathi–English;
Moroccan Arabic–English;
Norwegian–English; Punjabi–English;
Swahili–English; Swedish–English;
Vietnamese–English; Welsh–English
Eppler, E. 32, 99, 110
equivalence constraint 96–7
Ervin-Tripp S. 157
Espanglish 179
essentialism 11, 31
Estonian–English 156
ethnographic research 10, 14–16, 77
European Science Foundation 14
Fabbro, F. et al. 118, 124, 126, 128,
129, 130
Ferguson, C. 55
Finegan, E. 4n4
Finnis, K. 84–7
Finnish–English 157
Fishman, J. 55n3, 56, 179
“flagged” switches 107
Fletcher, P. 142
focus 19
Foley, J. A. 160
footing 68
Franceschini, R. 58
free morpheme constraint 95–6, 97, 146
French–Alsatian 1
children 156
constraints 95
grammar 108
social factors 43–8, 105, 106f
types of code-switching 37–9, 45–8, 110
French–Dutch (Brussels) 32, 37–8, 107, 110
French–Dutch–English 139–40
French–English 138–9
Canada 55, 61, 62t
children 148, 152, 153, 155
see also Jersey French
French–German 153, 158, 161
French–Italian 76n1
French–Kinyarwanda (Belgium) 70
French–Lingala 57–8, 168
French–Moroccan Arabic 24, 31, 81, 99
French–Portuguese 27
French–Swahili 57–8, 168
French–Tunisian Arabic 80, 81
functional grammar 94
functional head constraint 99, 119
fused lects 26–7
“fusion” 179
fuzziness 167–71
Gafaranga, J. 70, 74, 101, 141, 168
Gal, S. 42, 55
Galizan–Portuguese–Spanish 168
Gardner-Chloros, P. et al. 1, 2, 13, 14, 31, 32,
34, 37, 38, 39, 45, 46, 47, 57, 60, 64, 74–6,
78, 79–80, 81, 82–3, 84–7, 91, 95, 97, 105,
107, 108, 132
Garman, M. 135, 136, 140
Garrett, M. 125
Garrett, P. et al. 160
gender 18, 82–7, 172
accommodation 80–1
bonding 85–6
dampening directness 86–7
humour 85, 86–7
politeness 81, 83, 84–7
generative grammars 94
generativist approaches 95, 97–100
Government 97–8
“null” theories 98–100
Genesee, F. et al. 148, 149, 150, 153, 154, 155
Genovese, N. 21
George, A. 93
German–Dutch 148, 156
German–English 99, 110
Australia 108, 110, 143
children 148–9, 153, 158
German–French 153, 158, 161
German–Italian 155
German–Slovenian 66
German–Spanish 158
German–Turkish 113
gesture 176
Giacolone Ramat, Anna 20
Giancarli, P.-D. 150
Gibbons, J. 29, 81
Giesbers, H. 125, 134
Giles, H. 70–1
Goffman, E. 68
Goglia, F. 161
Goldblum, M. C. 126
Gomez, M. Y. 123
Goodz, N. 155
Gordon, M. 43, 66, 166
Government 97–8
grammar 10, 91–114, 172
and borrowing 31
Chomskyan/Universalist grammar 93
cognitive/functional/word grammars 94
CS’s “language-like” properties 107–9
formal grammars 94
future research 177
generativist approaches 95, 97–100
and lexicon 5
Matrix Language Frame (MLF) model 92,
100–4, 112, 137
Muysken’s “bilingual speech” 104–6
prescriptive/pedagogical grammar 92–3
production approaches 95, 100–4
ragged/paratactic/disjointed code-switching
sentences 91–2, 112
sociolinguistic v. typological factors 105,
“speech grammar” 112
Tree Adjoining Grammar 99
unconventional proposals 111–12
variationist approaches 95–7
Grammont’s Law 144
Greek Cypriots in London
attitudes 81–2
competence 2
constraints 96
gender and CS 83, 84–7
London Greek Radio 79
social factors 43–4, 48–54, 177
symbolic duality 52–4
types of code-switching 31, 34, 39, 49–52
Greek–English 136
Greek–Latin 88–90b
Greek–Turkish 23, 37, 63
Green, D. W. 122, 127, 129–32, 130f,
140–1, 174
Greenberg, J. H. 93
Grice, P. 69
Grosjean, F. et al. 118, 122, 124, 125, 127, 129,
136–40, 163, 174–5
Gumperz, J. J. 9, 20, 25, 27, 33, 54, 56–9, 66,
68, 77, 79, 81, 109, 165–6, 171
Gurindji–English 55
Gut, U. 153
Guttfreund, D. 124
Hakohol (newsletter) 21, 114–16b
Halmari, H. 110, 157, 170
Hamers, J. 35, 142, 144, 146, 147, 150, 158
Haugen, E. 9, 12
Haust, D. 76, 83
Heath, J. 31, 34
Hebrew–English 114–16b, 124
Hebrew–Yiddish 21
Heine, B. 37
Heller, M. 29–30, 55
Heritage, J. 184
Hernandez, E. 20, 54
Hewitt, R. 28
Hill, J. 77
Hill, K. 77
Hindi–English 78
Hindi–Punjabi (Delhi) 25
history of code-switching 4, 11, 13, 20, 24,
56, 178
Højrup, T. 56
Hudson, L. 170
Hulk, A. 158
humour 85, 86–7
Hungarian–English 152
Hurtado, N. M. 160
Huwaë, R. 110
Hyltenstam, K. 125
I-language 93
identity 5, 21, 27, 34, 37, 178–9 see also
sociolinguistic factors
Igbo-Nigerian–Italian 161
“immigrais” 27
“increment” 112
inhibition 130, 131–2, 141, 174
insertion 12, 105, 106f, 173
integration 12
intentionality 77
interference 9, 12, 30, 36, 146
interpreting 121n4, 140
Italian–English 158–9
Italian–French 76n1
Italian–German 155
Italian–Igbo-Nigerian 161
Italian–Sardinian 25, 104
Jackendoff, R. 135
Jacobson, R. 113
Jake, J. L. 102, 104, 108, 129, 135–6, 141,
Janicki, K. 10
Japanese 67–8b
Japanese–English 119, 151
Javier, R. A. 130
Jersey French (Jèrriais) 32–3
Jespersen, O. 35
Jisa, H. 157
Johanson, L. 37, 110–11
Johnson, C. 153
Johnson, J. S. 142
Jones, M. 20, 32–3, 95n2
Joshi, A. K. 95, 99, 100
Juan-Garau, M. 150, 156
Kamwangamalu, N. 119
Karousou-Fokas, R. 135, 136, 140
Kaufman, T. 36–7, 42, 171
Kecskes, I. 135
Kenya 160
Kessler, C. 151, 158–9
Kinyarwanda–French (Belgium) 70
Klavans, J. L. 100, 103
koineization 36
Kolers, P. A. 134
Köppe, R. 158
Korean–English 157
Kuteva, T. 37
L2 see second-language learners
Labov, W. 42
Lambert, W. 157
Lancaster, P. 153
language change 21, 23, 24–30, 113
language choices 13, 80
language contact 20–41
convergence v. preserving distinctiveness
21, 23
language interaction 30–6
language shift/change 21, 23, 24–30, 113
structural v. social influences 36–9
written texts 20–1, 22–3b, 40b
language death 26
language interaction 4, 30–6, 166, 202
borrowing 30–3
mixed languages 34–6
pidginization/creolization 3, 27–8, 33–4, 162
see also LIDES Project
language mixing 26–7
language modes 118, 136–40, 137f, 163
language shift 21, 23, 24–30
language tags 121
language(s) 8–9, 18–19, 65, 113, 118–19, 147,
173, 178
Lanvers, U. 157
Lanza, E. 144, 152–3, 155
Latin–Greek 88–90b
Latina (magazine) 21, 22–3b
Latvian–English 156, 158
Lavandera, B. 24
Lawson-Sako, S. 80, 81
laziness 14–15
Le Page, R. B. 8–9, 16, 18, 19, 33–4, 65, 82,
113, 170
Levelt, W. J. M. 120, 121, 132, 133f, 174
Levinson, S. 84
lexical hypothesis 121
lexicon 5
native synonym displacement 32–3
separate and joint access 120–1
Li, C. 119
Li Wei et al. 14, 55, 56, 70, 72–3, 109, 129, 166,
168, 169, 189–90
LIDES Project (Language Interaction Data
Exchange System) 9, 14, 60, 166, 177,
automated data analysis 198–200, 199t, 200t
CHAT 183, 184, 185–7, 188, 191–5, 197–8
CHILDES 182, 183–5
CLAN programs 183, 187, 198–9, 201
datasets available 200–1
glossary of terms 201–4
goals 182–7
language interaction data 197–8
minimal transcription process 195–7
research contacts 201
transcription 187–91
Unicode 184
Lin, A. M. Y. 159
Lindholm, K. A. 158
Lingala–French 57–8, 168
LIPPS Group see LIDES Project
Lipski, J. M. 96
Lisker, L. 139n8
loan-blending 146
local effects 72, 166
logogens 129
London Greek Radio 79
London Jamaican 3, 27, 28, 33–4, 57
Lüdi, G. 161
Lwidakho–Swahili 71–2
Ma’a 37
McClure, E. 32, 60–1
McClure, M. 60–1
McConvell, P. 55
McCormick, K. 81, 160, 169
McLelland, N. 20
MacSwan, J. 98, 139, 167
MacWhinney, B. 142, 182, 183, 190
Maehlum, B. 59
Mahootian, S. 21, 22–3b, 99
Mainville, S. 97, 110, 172
Malay varieties 159
Malay–Chinese 157
Malay–Dutch 110
Malay–English 159
Mandinka–Wolof–English 76
Maneva, B. 150
Mangruel, A. 179
Maniakas, T. M. 50
Marathi–English 111
Marathi–Sanskrit 111
Marcos, L. R. 130
Markedness Theory 56, 69–70, 171
Markovitch, J. 20–1
Martin, P. 159–60
Martin-Jones, M. 159
Marty, S. 125
Maters, K. 97
Matras, Y. 35, 132
Matrix Language 8, 46n1, 47, 92, 94, 110,
112–13, 173
Matrix Language Frame (MLF) model 100–4,
112, 137
grammatical criteria 103
morpheme-type criterion 102–3
psycholinguistic aspect 103
quantitative criterion 101–2
social aspect 103
use by other researchers 103–4
Matthews, S. 150
Mauritian Creole 162
Mayes, P. 177
medieval code-switching 24, 35–6, 40b
Meechan, M. 31
Meeuwis, M. 13, 57–8, 70, 141, 168
Meisel, J. M. 13, 146, 148, 149, 153, 156, 158
Melanesian Pidgin 27, 33
Merritt, M. et al. 160
metaphorical switching 59
methodological approaches 8, 18
Meuter, R. F. I. 121, 126, 141
Mexico 60–1, 83
Michif 35
Miles, F. 17
Miller, J. 139
Milroy, L. 9, 12, 43, 66, 153, 157, 166, 189–90
Minimalist Program 98, 99
Minnis, D. D. 154
Mishina, S. 153, 154
Mitchell, R. 17
MIX 29
mixed discourse 1
mixed languages 34–6
MLF see Matrix Language Frame (MLF) model
models of bilingual production 128–40, 129f,
De Bot 132–5, 174
Green 129–32, 130f, 140–1, 174
Grosjean 136–40, 174–5
Myers-Scotton and Jake 135–6
Moore, D. 48
Moreno, E. M. et al. 140
Morita, E. 151
Moroccan Arabic–Dutch 93, 98, 140
Moroccan Arabic–English 34
Moroccan Arabic–French 24, 31, 81, 99
morphophonemic integration 31–2
Morton, J. 129
Mougeon R. 30
Moyer, M. 135
Moyer, M. G. 103–4, 171, 185–6, 195, 199t,
Müller, N. 158
multilingualism 16
Muñoz, M. C. 27–8
Muysken, P. 9, 12, 46n1, 50, 60, 91, 93, 96, 98,
102, 103, 104–6, 106f, 107, 108, 110, 112,
154, 170, 201
Myers-Scotton, C. 26, 30, 46, 46n1, 56, 66,
69–70, 71–2, 74, 93, 94, 99, 100–1, 102,
104, 108, 113, 129, 135–6, 137, 141,
Nehrlich, B. 106
Neufeld, G. G. 127
neurolinguistics 118
Newport, E. L. 142
Nilep, C. 87
nonce borrowing 12
Nortier, J. 93, 98, 103, 170
Norwegian dialects 109
Norwegian–English 152, 153
“null” theories 98–100
Obler, L. K. et al. 121, 124, 126, 127
Ochs, E. 187
Oesch-Serra, C. 76n1
OPOL (One Parent One Language) 144, 150,
155, 162
Otheguy, R. 41
Padilla, A. M. 158
Pandharipande, R. V. 111
Pandit, I. 99
Papua New Guinea 33, 77
Paradis, J. 127, 148, 156
Paradis, M. 126, 134
paratactic code-switching 106–7
Park, T. 151, 158
Pauwels, A. 83
Pavlenko, A. 123, 124
Peal, E. 157
pedagogical grammar 92–3
Penfield, W. 127
Perecman, E. 125
Pérez-Vidal, C. 150, 156
Pfaff, C. 95, 96, 105, 113
phonology of code-switching 139, 178
pidginization 27, 33–4
plurilingualism 5, 7, 16–17, 18, 176, 179
politeness 81, 83, 84–7
bonding 85–6
dampening directness 86–7
humour 85, 86–7
Poplack, S. 12, 30, 31, 32, 61–3, 62t, 83, 93, 95,
96, 97, 104–5
Popper, K. 8
Portuguese–French 27
Portuguese–Spanish–Galizan 168
Pötzl, O. 126, 127
Poulisse, N. 134, 160–1
poverty of stimulus argument 154
Pragmatic approaches 7, 10 see also
conversational code-switching
Preference Organization 72
prescriptive grammar 92–3
Principles and Parameters theory 99
Proceso 61
production approaches 95, 100–4
psycholinguistic approaches 11–12, 117–41,
167, 172–5, 176
bilingual aphasia 122–6
brain localization/lateralization 126–8
encoding and production 15–16, 121–2,
experimental tasks 119–20
future research 177–8
“languages” 118–19
methodological issues 118–20
separate and joint access to lexicon 120–1
see also models of bilingual production
Pujolar i Cos, J. 41
Punjabi–Creole 58
Punjabi–English 31, 34, 39, 57, 74–6,
81, 83
Punjabi–Hindi (Delhi) 25
Quay, S. 153, 155, 175
questions, parenthetical 75
quotation 75–6, 176–7
Radford, A. 8
radio 79
Raggasonic Crew Feat 28–9
ragged code-switching 106–7
Rampton, B. 58, 168, 179
Rational Choice Model 69–70
reading 137–8
Redlinger, W. E. 151, 158
register 4, 4n4, 63
reiteration 75
relexification 24, 24n2
reliability 118
religious contexts 114–16b
repetition 75
Reyes, I. 157
Rindler-Schjerve, R. 25, 104
Ritchie, W. C. 96
Roberts, L. 127
Romaine, S. 5, 30, 33, 34, 81, 93, 96, 97,
144, 169
Romania 60
Rozensky, R.H. 123
Rubagumya, C. M. 160
Rubin, E. J. 119–20
Sachdev, I. 71, 80, 81
Sankoff, D. et al. 12, 31, 95, 97, 110,
112, 172
Sanskrit–Marathi 111
Sardinian–Italian 25, 104
Sarkar, M. 29
Sauris 25, 55
Saxe, J. G. vii–viii
Schelletter, C. 148
Schendl, H. 24, 40b
Schmidt, A. 24
Schreuder, R. 133, 134, 135
Scotton, C. M. 96
script-switching 21, 67, 67–8b, 114–16b
Seaman, P. D. 50
Sebastián-Gallés, N. 150
Sebba, M. 3, 27, 28, 57, 96, 113–14, 166, 168,
second-language learners 15, 17, 119, 125–6,
142, 159, 160–1, 175, 176
Sella-Mazi, E. 63
sentences 91–2, 112
Shin, S. J. 153, 157
Siegel, J. 33
Sifianou, M. 84
sign language 125
simultaneous interpreting 121n4, 140
Singh, R. 57
Sinka, I. 156, 158
situational code-switching 58–9
Slovenian–German 66
Smith, N. 18
Smith, W. 157
Smith, Z. 20, 28
Soares, C. 139–40
social networks 56, 168, 169, 171–2
social psychological aspects
accommodation 78–81, 118, 172
attitudes 14–16, 81–2, 171–2, 178–80
audience design 64, 78–81, 118
sociolinguistic factors 5, 10, 42–64, 166,
167, 171–2
comparisons between/within communities
future research 177
Greek Cypriots in London 43–4, 48–54, 177
macro-linguistic approaches 54–6
medieval code-switching 24, 35–6, 40b
methodology 8
situational code-switching 58–9
Strasbourg 43–8, 105, 106f
types of factor 42–3
v. structural factors 36–9
v. typological factors 105, 109–11
we-codes/they-codes 56–8, 66
see also conversational code-switching;
gender; social psychological aspects
song lyrics 28–9
South Africa 81
Spanish–English 119–20, 124, 140
children 150, 153, 158, 159
Espanglish 179
Spain 60
United States 21, 22–3b, 66, 85, 93, 95–6,
105, 113, 156–7, 179, 180
Spanish–Galizan–Portuguese 168 see also
Spanish–German 158
speaker awareness/education 111
speakers’ insights 14–16
“speech grammar” 112
Speech Production Model 133, 133f
Spolsky, B. 20
spontaneous translation 125
Sridhar, K. 13
Sridhar, S. 13
“stand alone” segments 111
Stroud, C. 77, 125, 176
structural v. social influences 36–9
studies of code-switching 9–10, 13–14
style 63–4 see also audience design
Swahili–English 69–70, 71–2, 96, 105, 106f
Swahili–French 57–8, 168
Swahili–Lwidakho 71–2
Swedish–English 157
Swigart, L. 58, 83
symbolism 66–7, 67–8b
Symposia on Bilingualism 14
syntax 8
Tabouret-Keller, A. 19, 33–4, 82, 113, 143,
156, 170
Taeschner, T. 151, 155
Tamis, A. 50
Tannen, D. 77, 177
television 78–9
Tenejapa 83
terminology 7, 10–13, 16, 146–7, 201–4
theoretical linguistics 8, 18
Thomason, S. G. 21, 23, 35, 36–7, 42,
152, 171
Thrace (Greece) 63
Timm, L. 20, 95
Tok Pisin 33, 77
Tolstoy, L. 20
Tonga 143
Toribio, A. J. 41, 119–20
Torras, M.-C. 101, 168, 189
Tracy, R. 8, 141, 147, 148–9, 154,
163–4, 169
transcription conventions xii see also LIDES
translinguistic wording 161
transversion 12, 17, 167, 178
Tree Adjoining Grammar 99
Treffers-Daller, J. 32, 37–8, 60, 83, 103, 107,
110, 170
triggering 140
trilingualism 16–17, 139–40
Trotter, D. A. 20
Trudgill, P. 24, 36, 42
Tsitsipis, L. D. 26, 37
Tunisian Arabic–French 80, 81
Turkish–Dutch 105, 108
Turkish–German 113
Turkish–Greek 23, 37, 63
typological factors 65, 109–11
Uniform Structure Principle 103
United States
Dutch–English 151–2
Spanish–English 21, 22–3b, 66, 85, 93, 95–6,
105, 113, 156–7, 179, 180
Universal Grammar (UG) 93
Urban Wolof (Dakar) 58
Valdes-Fallis, G. 20
validity 118
values 176
Van Eyck, Jan 21
Van Hell, J. 130, 139, 140, 174–5
variationist approaches 8, 95–7
clitic constraint 95
equivalence constraint 96–7
free morpheme constraint 95–6, 97, 146
“verbal action” 70–3, 172
Vietnamese–English 108
Vihman, M. M. 151, 154, 156
voice 75, 77, 176
Voice Onset Time (VOT) 139, 153
Volterra, V. 151
Von Raffler Engel, W. 176
Vygotsky, L. 157
Wasow, T. 167
we-codes and they-codes 56–8, 66
Weenik, D. 185
Wei Zhang 64
Weinreich, U. 8, 9
Welsh–English 104
Wernicke, C. 124, 126
Wilson, R. D. 33
Winer, L. 29
Winford, D. 103, 113, 169
Winter, J. 83
Wolof 58, 76
Woolard, K. A. 21, 55, 80–1, 108–9, 141, 169, 178
Woolford, E. 99
Wootton, T. 57, 166
word grammar 94
written texts with code-switching 20–1, 22–3b,
XML (Extensible Markup Language) 184, 204
Yiddish–Hebrew 21
Yip, V. 150
Zairian community, Belgium 57–8, 70, 168
Zee TV 78–9
Zentella, A. C. 81, 85, 143, 156–7, 180
Zimman, L. 32

Similar documents